A Review of the Event Analysis of Systemic Teamwork Methodology
A Review of the Event Analysis of Systemic Teamwork Methodology
A Review of the Event Analysis of Systemic Teamwork Methodology
- No tags were found...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Analysing C4i Activity: A <strong>Review</strong><br />
<strong>of</strong> <strong>the</strong> <strong>Event</strong> <strong>Analysis</strong> <strong>of</strong> <strong>Systemic</strong><br />
<strong>Teamwork</strong> (EAST) <strong>Methodology</strong><br />
The work described in this document has been undertaken by <strong>the</strong> Human Factors<br />
Integration Defence Technology Centre, part funded by <strong>the</strong> Human Capability Domain <strong>of</strong><br />
<strong>the</strong> U.K. Ministry <strong>of</strong> Defence Scientific Research Programme.<br />
© Human Factors Integration Defence Technology Centre 2005. The authors <strong>of</strong> this<br />
report have asserted <strong>the</strong>ir moral rights under <strong>the</strong> Copyright, Designs and Patents act,<br />
1988, to be identified as <strong>the</strong> authors <strong>of</strong> this work.<br />
Reference ............................................HFIDTC/WP1.1.3/10<br />
Version.................................................................................2<br />
Date.............................................................31 October 2005<br />
©Human Factors Integration Defence Technology Centre 2005
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Authors<br />
Paul Salmon<br />
Neville Stanton<br />
Dr Guy Walker<br />
Dan Jenkins<br />
Brunel University<br />
Brunel University<br />
Brunel University<br />
Brunel University<br />
ii
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Contents<br />
1 Executive Summary ................................................................................... 1<br />
2 Introduction ................................................................................................ 2<br />
3 Methods <strong>Review</strong>......................................................................................... 5<br />
3.1 EAST – <strong>Event</strong> <strong>Analysis</strong> <strong>of</strong> <strong>Systemic</strong> <strong>Teamwork</strong>.................................................................. 5<br />
3.2 Observation....................................................................................................................... 26<br />
3.3 HTA – Hierarchical Task <strong>Analysis</strong>..................................................................................... 34<br />
3.4 CDA – Co-Ordination Demands <strong>Analysis</strong> ......................................................................... 39<br />
3.5 CUD – Comms Usage Diagram........................................................................................ 47<br />
3.6 Social Network <strong>Analysis</strong>.................................................................................................... 52<br />
3.7 Operation Sequence Diagrams......................................................................................... 60<br />
3.8 Critical Decision Method ................................................................................................... 68<br />
3.9 Propositional Networks ..................................................................................................... 75<br />
4 Feedback from Questionnaire .................................................................. 87<br />
4.1 Aim .................................................................................................................................... 87<br />
4.2 Participants........................................................................................................................ 87<br />
4.3 Equipment ......................................................................................................................... 87<br />
4.4 Procedure.......................................................................................................................... 87<br />
4.5 Results .............................................................................................................................. 88<br />
4.6 Discussion <strong>of</strong> Results........................................................................................................ 95<br />
5 Summary.................................................................................................. 96<br />
6 Bibliography.............................................................................................. 99<br />
7 Appendix ................................................................................................ 102<br />
iii
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
1 Executive Summary<br />
This report describes a review <strong>of</strong> <strong>the</strong> <strong>Event</strong> <strong>Analysis</strong> <strong>of</strong> <strong>Systemic</strong> <strong>Teamwork</strong> (EAST)<br />
methodology (Baber and Stanton 2004) that was used to analyse C4i activity in <strong>the</strong> rail,<br />
air traffic control, energy distribution, military, police, fire service and naval domains.<br />
The EAST methodology and its component methods were analysed using a set <strong>of</strong> human<br />
factors (HF) methods and criteria designed to analyse <strong>the</strong> performance <strong>of</strong> <strong>the</strong> methods.<br />
The analysis was based upon <strong>the</strong> EAST applications undertaken as part <strong>of</strong> <strong>the</strong> Human<br />
Factors Integration Defence Technology Centre (HFI DTC) work programme (see<br />
application domains above).<br />
The methods review indicated that <strong>the</strong> EAST methodology is an exhaustive technique<br />
that is particularly suited to <strong>the</strong> analysis <strong>of</strong> C4i, team based and collaborative activity.<br />
The methods review also indicated that <strong>the</strong> outputs from <strong>the</strong> EAST methodology are<br />
extremely useful, <strong>of</strong>fering a number <strong>of</strong> different analyses <strong>of</strong>, and perspectives on <strong>the</strong><br />
scenario in question. Fur<strong>the</strong>r, it was concluded that <strong>the</strong> EAST methodology is relatively<br />
simple to learn and apply. In terms <strong>of</strong> flaws in <strong>the</strong> EAST methodology, <strong>the</strong> review<br />
indicated that <strong>the</strong> method can be overly time consuming (although this reflects <strong>the</strong><br />
complexity <strong>of</strong> C4i domains) in some instances (i.e. for larger, more complex tasks) and<br />
also that <strong>the</strong>re may be issues with <strong>the</strong> reliability <strong>of</strong> <strong>the</strong> method in some instances.<br />
1
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
2 Introduction<br />
The event analysis <strong>of</strong> systemic teamwork (EAST) methodology (Baber and Stanton 2004)<br />
was developed as part <strong>of</strong> <strong>the</strong> HFI DTC work programme, and uses a combination <strong>of</strong> HF<br />
methods to form a framework for analysing command, control, communication,<br />
computers and intelligence (C4i) activity. With a view to <strong>the</strong> development <strong>of</strong> a generic<br />
model <strong>of</strong> C4i, EAST has been applied to a number <strong>of</strong> diverse C4i scenarios, across a<br />
number <strong>of</strong> different domains. The domains and scenarios in which EAST has been used<br />
to analyse C4i activity are presented in Table 1.<br />
Table 1 – EAST <strong>Analysis</strong><br />
Domain<br />
Air Traffic Control<br />
National Air Traffic Services<br />
Energy Distribution<br />
National Grid Transco<br />
Fire Service<br />
Military Aviation<br />
E3D<br />
Navy<br />
HMS Driad<br />
Police<br />
Rail<br />
(Signalling)<br />
Scenario<br />
Holding<br />
Overflight<br />
Departure<br />
Approach<br />
Shift handover<br />
Barking switching operations<br />
Feckenham switching operations<br />
Tottenham return to service operations<br />
Alarm handling operations<br />
Chemical incident at remote farmhouse<br />
Road traffic accident involving chemical tanker<br />
Factory fire exercise #1<br />
Factory fire exercise #2<br />
General operation<br />
Air threat<br />
Surface threat<br />
Sub-surface threat<br />
Car break in caught on CCTV<br />
Suspected car break in<br />
Mobile phone robbery<br />
Detachment scenario<br />
Emergency Possession scenario<br />
Handback a Possession scenario<br />
Possession scenario<br />
EAST review<br />
In order to analyse <strong>the</strong> performance <strong>of</strong> <strong>the</strong> EAST methodology and its component<br />
methods, a review <strong>of</strong> <strong>the</strong> method was conducted based upon <strong>the</strong> applications described<br />
above. The review was based upon <strong>the</strong> same criteria that were used in <strong>the</strong> HF methods<br />
review (Salmon et al 2004a) that was conducted for <strong>the</strong> development <strong>of</strong> <strong>the</strong> EAST<br />
methodology. The criteria are outlined below:<br />
1. Name and acronym – <strong>the</strong> name <strong>of</strong> <strong>the</strong> technique and its associated acronym.<br />
2
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
2. Author(s), affiliations(s) and address(es) – <strong>the</strong> names, affiliations and addresses <strong>of</strong><br />
<strong>the</strong> authors are provided to assist with citation and requesting any fur<strong>the</strong>r help in<br />
using <strong>the</strong> technique.<br />
3. Background and applications – This section introduces <strong>the</strong> method, its origins and<br />
development, <strong>the</strong> domain <strong>of</strong> application <strong>of</strong> <strong>the</strong> method and also application areas<br />
that it has been used in.<br />
4. Domain <strong>of</strong> application – describes <strong>the</strong> domain that <strong>the</strong> technique was originally<br />
developed for and applied in.<br />
5. Procedure and advice – This section describes <strong>the</strong> procedure for applying <strong>the</strong><br />
method as well as general points <strong>of</strong> expert advice.<br />
6. Flowchart – A flowchart is provided, depicting <strong>the</strong> methods procedure.<br />
7. Advantages – Lists <strong>the</strong> advantages associated with using <strong>the</strong> method in <strong>the</strong> design<br />
<strong>of</strong> C4 systems.<br />
8. Disadvantages - Lists <strong>the</strong> disadvantages associated with using <strong>the</strong> method in <strong>the</strong><br />
design <strong>of</strong> C4 systems.<br />
9. Example – An example, or examples, <strong>of</strong> <strong>the</strong> application <strong>of</strong> <strong>the</strong> method are<br />
provided to show <strong>the</strong> methods output.<br />
10. Related methods – Any closely related methods are listed, including contributory<br />
and similar methods.<br />
11. Approximate training and application times - Estimates <strong>of</strong> <strong>the</strong> training and<br />
application times are provided to give <strong>the</strong> reader an idea <strong>of</strong> <strong>the</strong> commitment<br />
required when using <strong>the</strong> technique. Training and application times are classified<br />
as low (0 – 2 hours), medium (2 – 6 hours) or high (6 + hours).<br />
12. Reliability and Validity - Any evidence on <strong>the</strong> reliability or validity <strong>of</strong> <strong>the</strong> method<br />
are cited.<br />
13. Tools needed – Describes any additional tools required when using <strong>the</strong> method.<br />
14. Bibliography - A bibliography lists recommended fur<strong>the</strong>r reading on <strong>the</strong> method<br />
and <strong>the</strong> surrounding topic area.<br />
The criteria were applied to <strong>the</strong> overall EAST methodology and also <strong>the</strong> component<br />
individual methods. A summary <strong>of</strong> <strong>the</strong> EAST review is presented in Table 2. The review<br />
is presented in section 3 <strong>of</strong> this report.<br />
3
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 2 – Summary <strong>of</strong> EAST methods review<br />
Method<br />
<strong>Event</strong> <strong>Analysis</strong> <strong>of</strong><br />
<strong>Systemic</strong> <strong>Teamwork</strong><br />
(EAST)<br />
(Baber and Stanton<br />
2004)<br />
Observation<br />
Hierarchical Task<br />
<strong>Analysis</strong><br />
(Annett 2004)<br />
Co-ordination<br />
demands analysis<br />
(CDA)<br />
(Burke 2004)<br />
Type <strong>of</strong><br />
method<br />
Team<br />
analysis<br />
method<br />
Data<br />
collection<br />
Task<br />
analysis<br />
Team<br />
analysis<br />
Domain SME’s Related<br />
methods<br />
Generic Yes Obs<br />
C4i<br />
HTA<br />
CDA<br />
OSD<br />
CUD<br />
SNA<br />
CDM<br />
Prop Nets<br />
Training<br />
time<br />
Application<br />
time<br />
Tools<br />
needed<br />
High High Video /<br />
Audio<br />
recording<br />
equipment<br />
MS Word<br />
MS Excel<br />
MS Visio<br />
AGNA<br />
Generic No HTA Low High Video /<br />
Audio<br />
recording<br />
equipment<br />
MS Word<br />
Generic No Obs Med High Pen and<br />
paper<br />
MS Notepad<br />
Generic Yes Obs<br />
HTA<br />
Low Med Pen and<br />
paper<br />
MS Excel<br />
Reliability Validity Advantages Disadvantages<br />
Med High 1. EAST <strong>of</strong>fers an extremely<br />
exhaustive analysis <strong>of</strong> <strong>the</strong> C4i<br />
domain in question.<br />
2. EAST is relatively easy to train<br />
and apply. The provision <strong>of</strong> <strong>the</strong><br />
WESTT and AGNA s<strong>of</strong>tware<br />
packages also reduces application<br />
time considerably.<br />
3. The EAST output is extremely<br />
useful, <strong>of</strong>fering a number <strong>of</strong><br />
different analyses and<br />
perspectives on <strong>the</strong> C4i activity in<br />
question.<br />
High High 1. Acts as <strong>the</strong> primary input <strong>the</strong><br />
EAST methodology.<br />
2. Easy to conduct provided <strong>the</strong><br />
appropriate planning has been<br />
made.<br />
3. Allows <strong>the</strong> analysts to gain a<br />
deeper understanding <strong>of</strong> <strong>the</strong><br />
domain and scenario under<br />
analysis.<br />
Med High 1. Easy to learn and apply.<br />
2. Allows <strong>the</strong> analysts to gain a<br />
deeper understanding <strong>of</strong> <strong>the</strong><br />
domain and scenario under<br />
analysis.<br />
3. Describes <strong>the</strong> task under<br />
analysis in terms <strong>of</strong> component<br />
task steps and operations.<br />
Low High 1. Offers an overall rating <strong>of</strong> coordination<br />
between team members<br />
for each teamwork based task step<br />
in <strong>the</strong> scenario under analysis.<br />
2. Also <strong>of</strong>fers a rating for each<br />
teamwork behaviour in <strong>the</strong> CDA<br />
teamwork taxonomy.<br />
3. The technique can be used to<br />
identify task-work (individual) and<br />
team-work task steps involved in<br />
<strong>the</strong> scenario in question.<br />
1. Due to its exhaustive nature, EAST is<br />
time consuming to apply.<br />
2. Reliability may be questionable in some<br />
areas. Reliability <strong>of</strong> <strong>the</strong> method is still<br />
being established.<br />
3. A large portion <strong>of</strong> <strong>the</strong> output is<br />
descriptive. Great onus is placed upon <strong>the</strong><br />
analyst to interpret <strong>the</strong> results accordingly.<br />
1. Observations are typically time<br />
consuming to conduct (including <strong>the</strong><br />
lengthy data analysis procedure).<br />
2. It is <strong>of</strong>ten difficult to gain <strong>the</strong> required<br />
access to <strong>the</strong> establishment under<br />
analysis.<br />
3. There is no guarantee that <strong>the</strong> required<br />
data will be obtained.<br />
1. Can be difficult and time consuming to<br />
conduct for large, complex scenarios.<br />
2. Reliability is questionable. Different<br />
analysts may produce different HTA<br />
outputs for <strong>the</strong> same scenario.<br />
3. Provides mainly descriptive ra<strong>the</strong>r than<br />
analytical information. Also does not cater<br />
for <strong>the</strong> cognitive components <strong>of</strong> task<br />
performance.<br />
1. The CDA procedure can be time<br />
consuming and laborious. For each<br />
individual task step, seven teamwork<br />
behaviours are rated.<br />
2. To ensure validity, SME’s are required.<br />
3. Intra and inter-analyst reliability may be<br />
questionable.<br />
4
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
3 Methods <strong>Review</strong><br />
3.1 EAST – <strong>Event</strong> <strong>Analysis</strong> <strong>of</strong> <strong>Systemic</strong> <strong>Teamwork</strong><br />
Background and applications<br />
The EAST methodology (Baber and Stanton 2004) was developed for <strong>the</strong> analysis <strong>of</strong> C4i<br />
(command, control, communications, computers and intelligence) activity. EAST uses a<br />
combination <strong>of</strong> HF methods to form a framework for analysing C4i activity. A brief<br />
description <strong>of</strong> each component method is provided below:<br />
• Hierarchical Task <strong>Analysis</strong> (HTA) – involves describing <strong>the</strong> scenario under<br />
analysis using a hierarchy <strong>of</strong> goals, sub-goals and operations.<br />
• Observation – is used during an EAST analysis to ga<strong>the</strong>r data surrounding<br />
<strong>the</strong> scenario under analysis. The observational data obtained is used as <strong>the</strong><br />
primary input to an EAST analysis.<br />
• Co-ordination demands analysis (CDA) – involves defining <strong>the</strong> task-work<br />
and team-work tasks involved during <strong>the</strong> scenario, and <strong>the</strong>n rating each<br />
teamwork task step against <strong>the</strong> CDA taxonomy <strong>of</strong>; communication;<br />
situational awareness; decision making; mission analysis; leadership;<br />
adaptability; and assertiveness.<br />
• Comms Usage Diagram (CUD) - is used to describe collaborative activity<br />
between distributed agents. The output <strong>of</strong> CUD describes how and why<br />
communications between a team occur, which technology is involved in <strong>the</strong><br />
communication, and <strong>the</strong> advantages and disadvantages <strong>of</strong> <strong>the</strong> technology<br />
medium used.<br />
• Social Network <strong>Analysis</strong> – involves defining and analysing <strong>the</strong><br />
relationships between agents within a scenario network. A matrix <strong>of</strong><br />
association and a social network diagram are constructed, and agent<br />
centrality, sociometric status and network density figures are calculated.<br />
• Operation Sequence Diagram (OSD) – is used to represents <strong>the</strong> tasks, <strong>the</strong><br />
actors, <strong>the</strong> communications, <strong>the</strong> social organisation, <strong>the</strong> sequence and time<br />
in which <strong>the</strong> scenario took place. An OSD captures <strong>the</strong> flow <strong>of</strong> information<br />
among distributed actors and shows how this is mediated through<br />
technology and team working. Additionally, operational loafing figures are<br />
calculated for <strong>the</strong> following OSD operators; Operation, Receive, Delay,<br />
Decision, Transport, Combined Operations.<br />
• Critical Decision Method (CDM) – involves <strong>the</strong> use <strong>of</strong> interview probes to<br />
elicit information regarding agent decision making strategies adopted<br />
during <strong>the</strong> scenario under analysis.<br />
5
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
• Propositional Networks – The CDM output is used to construct<br />
propositional networks for each incident phase. Propositional networks are<br />
comprised <strong>of</strong> nodes that represent sources <strong>of</strong> information, agents, and<br />
objects that are linked through specific causal paths. The propositional<br />
network thus represents <strong>the</strong> ‘ideal’ collection <strong>of</strong> knowledge for an incident.<br />
The EAST methodology has been applied in a number <strong>of</strong> domains, including <strong>the</strong> fire<br />
service (Baber et al 2004a), police service (Baber et al 2004b) naval warfare (Stewart et<br />
al 2004), military aviation, energy distribution (Salmon et al 2004b), air traffic control<br />
and rail domains (Walker et al 2004), in order to analyse C4i activity. EAST is a<br />
comprehensive technique <strong>of</strong>fering a multi-faceted assessment <strong>of</strong> <strong>the</strong> C4i network in<br />
question. EAST provides an assessment <strong>of</strong> agent roles within <strong>the</strong> network, a description<br />
<strong>of</strong> <strong>the</strong> activity including <strong>the</strong> flow <strong>of</strong> information, <strong>the</strong> component tasks, communication<br />
between agents and <strong>the</strong> operational loading <strong>of</strong> each agent. Co-ordination between agents<br />
is also rated and <strong>the</strong> knowledge required throughout <strong>the</strong> task under analysis is defined.<br />
Domain <strong>of</strong> application<br />
Generic. The EAST methodology can be used in any domain that utilises a C4i<br />
infrastructure during task performance.<br />
Procedure and advice<br />
Step 1: Define scenario(s)<br />
The first step in an EAST analysis involves defining <strong>the</strong> scenario(s) that are to be <strong>the</strong><br />
subject <strong>of</strong> <strong>the</strong> analysis. It is recommended that this is done in collaboration with a SME<br />
from <strong>the</strong> domain in question. The chosen scenario(s) should be representative <strong>of</strong> all C4i<br />
activity in <strong>the</strong> domain in question. For example, in an analysis <strong>of</strong> C4i activity in <strong>the</strong> civil<br />
energy distribution (Salmon et al 2004), a switching out <strong>of</strong> circuit’s scenario and a return<br />
to service <strong>of</strong> circuits scenario were used.<br />
Step 2: Conduct HTA for <strong>the</strong> scenario(s) under analysis<br />
Once <strong>the</strong>y are clearly defined, <strong>the</strong> analyst(s) should conduct a HTA for each scenario.<br />
Again, it is recommended that this is done in collaboration with a relevant SME.<br />
Step 3: Plan observation<br />
The next step <strong>of</strong> <strong>the</strong> EAST analysis involves planning <strong>the</strong> observation <strong>of</strong> <strong>the</strong> scenario(s)<br />
under analysis. Typically <strong>the</strong> scenarios involved in C4i activity are dispersed over a<br />
number <strong>of</strong> different locations involving a number <strong>of</strong> agents, and so who to observe and<br />
where to observe it requires careful thought. Again, input from an appropriate SME is<br />
useful here. It is also useful at this stage to clearly define what data is to be collected. The<br />
use <strong>of</strong> recording equipment (e.g. type, location, focus etc) should also be clarified. If time<br />
permits, it may also be pertinent to conduct a trial run <strong>of</strong> <strong>the</strong> planned observation.<br />
Step 4: Observe scenario(s)<br />
6
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
The observation step is <strong>the</strong> most important part <strong>of</strong> <strong>the</strong> EAST procedure. Typically, a<br />
number <strong>of</strong> analyst(s) are used in scenario observation. All activity involved in <strong>the</strong><br />
scenario under analysis should be recorded along an incident timeline, including a<br />
description <strong>of</strong> <strong>the</strong> activity, <strong>the</strong> agents involved, any communications made and <strong>the</strong><br />
technology involved. Additional notes should be made where required, including <strong>the</strong><br />
purpose <strong>of</strong> <strong>the</strong> activity, any errors made and also any information that <strong>the</strong> agent involved<br />
feels is relevant.<br />
Step 5: Critical decision method<br />
Once <strong>the</strong> scenario under analysis is complete, each ‘key’ (e.g. scenario commander,<br />
agents performing <strong>the</strong> component tasks etc) involved should be subjected to a CDM<br />
interview. This involves dividing <strong>the</strong> scenario into key incident phases and <strong>the</strong>n<br />
interviewing <strong>the</strong> agent involved using pre-defined probes. A CDM interview should be<br />
conducted for each incident phase.<br />
Step 6: Transcribe scenario data<br />
Once all <strong>of</strong> <strong>the</strong> scenario data is collected, it should be transcribed in order to make it<br />
compatible to <strong>the</strong> EAST analysis phase. An event transcript should be constructed. The<br />
transcript should describe <strong>the</strong> scenario over a timeline, including descriptions <strong>of</strong> activity,<br />
<strong>the</strong> agents involved, any communications made and <strong>the</strong> technology used. In order to<br />
ensure <strong>the</strong> validity <strong>of</strong> <strong>the</strong> data, <strong>the</strong> scenario transcript should be reviewed by one <strong>of</strong> <strong>the</strong><br />
SME’s involved.<br />
Step 7: Re-iterate HTA<br />
The data transcription process allows <strong>the</strong> analyst(s) to gain a deeper and more accurate<br />
understanding <strong>of</strong> <strong>the</strong> scenario under analysis. It also allows any discrepancies between<br />
<strong>the</strong> initial HTA scenario description and <strong>the</strong> actual activity observed to be resolved.<br />
Typically, C4i activity does not run entirely according to protocol, and certain tasks may<br />
have been performed during <strong>the</strong> scenario that were not described in <strong>the</strong> initial HTA<br />
description. The analyst should compare <strong>the</strong> scenario transcript to <strong>the</strong> initial HTA, and<br />
add any changes as required.<br />
Step 8: Co-ordination demands analysis<br />
The CDA involves extracting teamwork tasks from <strong>the</strong> HTA and rating <strong>the</strong>m against <strong>the</strong><br />
associated CDA taxonomy. Each teamwork task is rated against each CDA behaviour on<br />
a scale <strong>of</strong> 1 (low) to 3 (high). Total co-ordination for each teamwork step can be derived<br />
by calculating <strong>the</strong> mean across <strong>the</strong> CDA behaviours. The mean total co-ordination figure<br />
for <strong>the</strong> scenario under analysis should also be calculated.<br />
Step 9: Construct comms usage diagram<br />
The CUD is used to represent communication between <strong>the</strong> agents involved in <strong>the</strong> scenario<br />
and also to analyse <strong>the</strong> comms technology utilised. The scenario transcript is used as <strong>the</strong><br />
input to <strong>the</strong> CUD.<br />
Step 10: Conduct social network analysis<br />
7
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
The SNA is used to analyse <strong>the</strong> relationships between <strong>the</strong> agents involved in <strong>the</strong> scenario<br />
under analysis. Using <strong>the</strong> scenario transcript, <strong>the</strong> analyst should firstly construct an agent<br />
association matrix, which presents <strong>the</strong> links between <strong>the</strong> agents and also <strong>the</strong> frequency <strong>of</strong><br />
communications between <strong>the</strong> agents. An example association matrix is presented in Table<br />
6. From <strong>the</strong> association matrix, a social network diagram is constructed and agent<br />
centrality, sociometric status, and network density are calculated. It is recommended that<br />
<strong>the</strong> Agna SNA s<strong>of</strong>tware package is used for <strong>the</strong> SNA phase <strong>of</strong> <strong>the</strong> EAST methodology.<br />
Step 11: Construct operation sequence diagram<br />
The OSD represents <strong>the</strong> activity observed during <strong>the</strong> scenario under analysis. The analyst<br />
should construct <strong>the</strong> OSD using <strong>the</strong> scenario transcript and <strong>the</strong> associated HTA as inputs.<br />
Once <strong>the</strong> initial OSD is completed, <strong>the</strong> analyst should <strong>the</strong>n add <strong>the</strong> results <strong>of</strong> <strong>the</strong> CDA to<br />
each teamwork task step.<br />
Step 12: Construct Propositional networks<br />
The final step <strong>of</strong> <strong>the</strong> EAST analysis involves constructing propositional networks for<br />
each scenario phase identified during <strong>the</strong> CDM interviews. In order to construct <strong>the</strong><br />
propositional networks for each phase, a basic contents analysis should be conducted<br />
using <strong>the</strong> associated CDM outputs. Knowledge objects are defined as knowledge,<br />
information, artefacts and actions. Each knowledge object should have a corresponding<br />
node in <strong>the</strong> propositional network. Next, <strong>the</strong> links between <strong>the</strong> nodes should be specified.<br />
To do this, <strong>the</strong> analyst should specify any links between <strong>the</strong> knowledge objects within <strong>the</strong><br />
propositional networks, using <strong>the</strong> following links taxonomy;<br />
Example<br />
• Has<br />
• Is<br />
• Causes<br />
• Knows<br />
• Requires<br />
• Prevents<br />
The following example is taken from an analysis <strong>of</strong> a switching scenario drawn from <strong>the</strong><br />
civil energy distribution domain (Salmon et al 2004b). The scenario took place at Barking<br />
275Kv, 132Kv and 33Kv substations and <strong>the</strong> National Grid Transco (NGT) Network<br />
Operations Centre (NOC) in Warwick. The scenario under analysis involved <strong>the</strong><br />
switching out <strong>of</strong> three circuits (SGT5 and SGT1A and 1B) at Barking 275Kv, 132Kv and<br />
33Kv Substations. Circuit SGT5 was being switched out for <strong>the</strong> installation <strong>of</strong> a new<br />
transformer for <strong>the</strong> nearby channel tunnel rail link and SGT1A and 1B were being<br />
switched out for substation maintenance. The observation focussed upon four main<br />
8
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
agents involved in <strong>the</strong> scenario. A description <strong>of</strong> <strong>the</strong> agents involved is presented in Table<br />
3.<br />
Table 3 – Agents involved in Switching Scenario<br />
Agent Title<br />
NOC<br />
SAP<br />
WOK<br />
REC<br />
Details<br />
Control room operator based at <strong>the</strong> NOC<br />
Senior Authorised Person (SAP) at Barking substations.<br />
Also working with authorised person (AP)<br />
Control room operator based at Wokingham control room<br />
Control room operator based at <strong>the</strong> regional electricity<br />
company control room<br />
Activity was observed by two observers. The first observer was situated at <strong>the</strong> Network<br />
operations centre (NOC) at National Grid Transco (NGT) house in Warwick, and<br />
observed <strong>the</strong> activity <strong>of</strong> <strong>the</strong> NOC operator. The second observer was situated at Barking<br />
275Kv, 132Kv and 33Kv Substations, and observed <strong>the</strong> activity <strong>of</strong> <strong>the</strong> SAP and AP. The<br />
results <strong>of</strong> <strong>the</strong> CDA are presented in Table 4. An extract <strong>of</strong> <strong>the</strong> CDA is presented in Table<br />
5.<br />
Table 4 – Switching scenario CDA results<br />
Category<br />
Result<br />
Total task steps 314<br />
Total taskwork 114 (36%)<br />
Total teamwork 200 (64%)<br />
Mean Total Co-ordination 1.57<br />
Modal Total Co-ordination 1.00<br />
Minimum Co-ordination 1.00<br />
Maximum Co-ordination 2.14<br />
The agent association matrix is presented in Table 6. From <strong>the</strong> association matrix, a<br />
social network diagram was constructed (Figure 3-1). Directional arrows represent <strong>the</strong><br />
communications between agents, and <strong>the</strong> frequency <strong>of</strong> communications between agents is<br />
also presented. Centrality, and sociometric status were also calculated for each agent<br />
involved in <strong>the</strong> scenario. The SNA results are presented in Table 7.<br />
9
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
NOC<br />
Operator<br />
8<br />
2<br />
3<br />
1<br />
3<br />
SAP at<br />
Barking<br />
substation<br />
3<br />
WOK<br />
Operator<br />
2<br />
1<br />
2<br />
1<br />
1<br />
REC<br />
operator<br />
Figure 3-1 – Social Network Diagram<br />
10
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 5 – Extract <strong>of</strong> CDA analysis<br />
Task<br />
Step<br />
Agent Step No. Task<br />
Work<br />
Team<br />
Work<br />
Comm SA DM MA Lead Ad Ass TOT<br />
CO-<br />
ORD<br />
Mode<br />
TOT<br />
CO-<br />
ORD<br />
Mean<br />
1.1.1 WOK control room operator Use phone to contact NOC 1<br />
1.1.2 WOK control room operator Exchange identities 1 3 3 1 1 1 1 1 1.00 1.57<br />
NOC control room operator<br />
1.1.3 WOK control room operator Agree SSC documentation 1 3 3 3 1 1 1 1 1.00 1.86<br />
NOC control room operator<br />
1.1.4.1 NOC control room operator Agree SSC with WOK 1 3 3 3 1 1 1 1 1.00 1.86<br />
1.1.4.2 NOC control room operator Agree time with WOK 1 3 3 3 1 1 1 1 1.00 1.86<br />
1.1.5.1 NOC control room operator Record details onto log sheet 1<br />
1.1.5.2 NOC control room operator Enter details into WorkSafe 1<br />
1.2.1 NOC control room operator Ask for isolators to be opened<br />
1 3 3 1 2 2 1 1 1.00 1.86<br />
remotely<br />
1.2.2 WOK control room operator Perform remote isolation 1<br />
1.2.3 NOC control room operator Check barking s/s screen 1<br />
1.2.4 WOK control room operator End communications 1 3 1 1 1 1 1 1 1.00 1.29<br />
NOC control room operator<br />
1.3.1 NOC control room operator Use phone to contact SAP at 1<br />
Barking<br />
1.3.2 NOC control room operator Exchange identities 1 3 3 1 1 1 1 1 1.00 1.57<br />
SAP at Barking<br />
11
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 6 – Agent Association matrix<br />
A B C D<br />
A 0 3 2 1<br />
B 8 0 2 2<br />
C 3 3 0 0<br />
D 1 1 1 0<br />
Table 7 – SNA results<br />
Agent<br />
B-L<br />
Centrality<br />
Sociometric<br />
Status<br />
SAP at Barking 2.16 6.33<br />
NOC operator 2.16 6.00<br />
REC operator 1.85 2.00<br />
WOK control operator 1.85 3.66<br />
A CUD was constructed in order to analyse <strong>the</strong> communications made and <strong>the</strong> comms<br />
technology used. An extract <strong>of</strong> <strong>the</strong> CUD for <strong>the</strong> switching scenario is presented in Figure<br />
3-2.<br />
NGC 16 th June 2004 Scenario: Switching Operations<br />
Time<br />
NOC Control Room<br />
Operator<br />
SAP/AP at Barking<br />
Substation<br />
WOK Control<br />
Room Operator<br />
REC Control Room<br />
Operator<br />
Comms<br />
Effects <strong>of</strong> comms<br />
medium used<br />
Recommended comms<br />
11:55<br />
Contact NOC to<br />
agree SSC and<br />
transfer control to<br />
NOC<br />
+ Sound is clear and stable<br />
+ Comms is two way<br />
+ Can look at site diagram/desktop<br />
screen at <strong>the</strong> same time<br />
+ Confidentiality maintained<br />
- Data is not recorded. Information<br />
may be lost<br />
11:56<br />
Contact REC/EDF to<br />
confirm busbar<br />
reactor B operations<br />
complete<br />
+ Ensures high Mobility<br />
+ Sound is stable and clear<br />
+ Comms is two way<br />
+ Operator has hands free for<br />
o<strong>the</strong>r tasks<br />
- Battery may run out<br />
- Comms may be affected by poor<br />
signal<br />
- Data is not recorded, information<br />
may be lost<br />
11:57 Contact REC to give<br />
instructions regarding<br />
substation circuit and<br />
POI’s<br />
+ Sound is clear and stable<br />
+ Comms is two way<br />
+ Can look at site diagram/desktop<br />
screen at <strong>the</strong> same time<br />
+ Confidentiality maintained<br />
- Data is not recorded. Information<br />
may be lost<br />
12:17<br />
Contact SAP at<br />
Barking to give<br />
switching instructions<br />
+ Sound is clear and stable<br />
+ Comms is two way<br />
+ Can look at site diagram/desktop<br />
screen at <strong>the</strong> same time<br />
+ Confidentiality maintained<br />
- Data is not recorded. Information<br />
may be lost<br />
Figure 3-2 – CUD extract<br />
12
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
An OSD was constructed in order to represent <strong>the</strong> flow <strong>of</strong> activity during <strong>the</strong> switching<br />
scenario. The OSD glossary is presented in Figure 3-3. An extract <strong>of</strong> <strong>the</strong> OSD for <strong>the</strong><br />
switching scenario is presented in Figure 3-4. From <strong>the</strong> OSD, operational loading figures<br />
were calculated in order to determine agent loading during <strong>the</strong> scenario. The operational<br />
loading figures are presented in Table 8.<br />
Table 8 – Operational loading results<br />
Agent Operation Receive Transport Decision Delay Total<br />
NOC 98 40 138<br />
SAP 223 21 19 1 264<br />
WOK 40 10 50<br />
REC 15 14 29<br />
The operational loading figures indicate that <strong>the</strong> NOC operator was loaded <strong>the</strong> highest for<br />
receive (i.e. receipt <strong>of</strong> information) operations, and <strong>the</strong> SAP at Barking was loaded <strong>the</strong><br />
highest for operation, transport and delay operations.<br />
Figure 3-3 – OSD Glossary<br />
13
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Figure 3-4 – OSD Extract<br />
The CDM analysis for <strong>the</strong> switching scenario involved interviewing <strong>the</strong> SAP at Barking<br />
and <strong>the</strong> NOC control room operator. In both cases <strong>the</strong> probes defined by O’Hare et al.<br />
(2000) were used to evaluate <strong>the</strong> decision making during <strong>the</strong> scenario. The CDM probes<br />
used are presented in Table 9. For <strong>the</strong> purposes <strong>of</strong> <strong>the</strong> CDM analysis, <strong>the</strong> scenario was<br />
divided into four phases; First issue <strong>of</strong> instructions, deal with switching requests, perform<br />
isolation and report back to NOC. The CDM analysis is presented in Table 10 to Table<br />
13.<br />
14
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 9 – CDM probes (O’Hare et al 2000)<br />
Goal Specification<br />
Cue Identification<br />
Expectancy<br />
Conceptual<br />
Influence <strong>of</strong> uncertainty<br />
Information integration<br />
Situation Awareness<br />
Situation Assessment<br />
Options<br />
Decision blocking - stress<br />
Basis <strong>of</strong> choice<br />
Analogy/<br />
generalisation<br />
What were your specific goals at <strong>the</strong> various decision<br />
points<br />
What features were you looking for when you formulated<br />
your decision<br />
How did you that you needed to make <strong>the</strong> decision<br />
How did you know when to make <strong>the</strong> decision<br />
Were you expecting to make this sort <strong>of</strong> decision during <strong>the</strong><br />
course <strong>of</strong> <strong>the</strong> event<br />
Describe how this affected your decision making process.<br />
Are <strong>the</strong>re any situations in which your decision would have<br />
turned out differently<br />
Describe <strong>the</strong> nature <strong>of</strong> <strong>the</strong>se situations and <strong>the</strong><br />
characteristics that would have changed <strong>the</strong> outcome <strong>of</strong><br />
your decision.<br />
At any stage, were you uncertain about ei<strong>the</strong>r <strong>the</strong> reliability<br />
<strong>of</strong> <strong>the</strong> relevance <strong>of</strong> <strong>the</strong> information that you had available<br />
At any stage, were you uncertain about <strong>the</strong> appropriateness<br />
<strong>of</strong> <strong>the</strong> decision<br />
What was <strong>the</strong> most important piece <strong>of</strong> information that you<br />
used to formulate <strong>the</strong> decision<br />
What information did you have available to you at <strong>the</strong> time <strong>of</strong><br />
<strong>the</strong> decision<br />
Did you use all <strong>of</strong> <strong>the</strong> information available to you when<br />
formulating <strong>the</strong> decision<br />
Was <strong>the</strong>re any additional information that you might have<br />
used to assist in <strong>the</strong> formulation <strong>of</strong> <strong>the</strong> decision<br />
Were <strong>the</strong>re any o<strong>the</strong>r alternatives available to you o<strong>the</strong>r<br />
than <strong>the</strong> decision you made<br />
Was <strong>the</strong>ir any stage during <strong>the</strong> decision making process in<br />
which you found it difficult to process and integrate <strong>the</strong><br />
information available<br />
Describe precisely <strong>the</strong> nature <strong>of</strong> <strong>the</strong> situation<br />
Do you think that you could develop a rule, based on your<br />
experience, which could assist ano<strong>the</strong>r person to make <strong>the</strong><br />
same decision successfully<br />
Why/Why not<br />
Were you at any time, reminded <strong>of</strong> previous experiences in<br />
which a similar decision was made<br />
Were you at any time, reminded <strong>of</strong> previous experiences in<br />
which a different decision was made<br />
15
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 10 – CDM Phase 1: First issue <strong>of</strong> instructions<br />
Goal Specification<br />
Cue identification<br />
Expectancy<br />
Conceptual Model<br />
Uncertainty<br />
Information<br />
Situation Awareness<br />
Situation<br />
Assessment<br />
Options<br />
Stress<br />
Choice<br />
Analogy<br />
Establish what isolation <strong>the</strong> SAP at Barking is looking for. Depends on<br />
gear<br />
Don’t Believe It (DBI) alarm is unusual – faulty contact (not open or<br />
closed) questionable data from site checking rating <strong>of</strong> earth switches<br />
(may be not fully rated for circuit current – so additional earths may be<br />
required.<br />
Check that SAP is happy with instructions as not normal.<br />
Decision expected by DBI is not common.<br />
Recognised instruction but not stated in WE1000 – as <strong>the</strong>re are not<br />
too many front and rear shutters metal clad switch gear.<br />
Confirm from field about planned instruction – make sure that SAP is<br />
happy with <strong>the</strong> instruction.<br />
Reference to front and rear busbars.<br />
WE1000 procedure<br />
Metal clad switchgear<br />
Barking SGT1A/1B substation screen<br />
SAP at Barking<br />
Ask colleagues if needed to<br />
No alternatives<br />
N/A<br />
WE1000 – need to remove what does not apply<br />
Could add front and rear busbar procedures<br />
Best practice guide for metal clad EMS switching<br />
16
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 11 - CDM Phase 2: Deal with switching requests<br />
Goal Specification<br />
Cue identification<br />
Expectancy<br />
Conceptual Model<br />
Uncertainty<br />
Information<br />
Situation<br />
Awareness<br />
Situation<br />
Assessment<br />
Options<br />
Stress<br />
Choice<br />
Analogy<br />
Obtain confirmation from NOC that planned isolation is still required.<br />
Approaching time for planned isolation.<br />
Switching phone rings throughout building.<br />
Airblast circuit breakers (accompanied by sirens) can be heard to<br />
operate remotely (more so in Barking 275 than Barking C 132).<br />
Yes – routine planned work according to fixed procedures.<br />
Wokingham have performed remote isolations already.<br />
Circuit configured ready for local isolation.<br />
Physical verification <strong>of</strong> apparatus always required (DBI – don’t<br />
believe it).<br />
Proceduralised information from NOC – circuit, location, time, actions<br />
required etc.<br />
Switching log.<br />
Switching log.<br />
Physical status <strong>of</strong> apparatus.<br />
Planning documentation.<br />
Visual or verbal information from substation personnel.<br />
Planning documentation used only occasionally<br />
Refusal <strong>of</strong> switching request.<br />
Additional conditions to switching request.<br />
Some time pressure.<br />
Yes – highly proceduralised anyway<br />
Yes – routine activity<br />
17
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 12 – CDM Phase 3: Perform isolation<br />
Goal Specification<br />
Cue identification<br />
Expectancy<br />
Conceptual Model<br />
Uncertainty<br />
Information<br />
Situation Awareness<br />
Situation<br />
Assessment<br />
Options<br />
Stress<br />
Choice<br />
Analogy<br />
Ensure it is safe to perform local isolation.<br />
Confirm circuits/equipment to be operated.<br />
Telecontrol displays/circuit loadings.<br />
Equipment labels.<br />
Equipment displays.<br />
O<strong>the</strong>r temporary notices.<br />
Equipment configured according to planned circuit switching.<br />
Equipment will function correctly.<br />
Layout/type/characteristics <strong>of</strong> circuit.<br />
Circuit loadings/balance.<br />
Function <strong>of</strong> equipment.<br />
Will equipment physically work as expected (will something jam etc.).<br />
O<strong>the</strong>r work being carried out by o<strong>the</strong>r parties (e.g. EDF).<br />
Switching log.<br />
Visual and verbal information from those undertaking <strong>the</strong> work.<br />
Physical information from apparatus and telecontrol displays.<br />
All information used<br />
Inform NOC that isolation cannot be performed/o<strong>the</strong>r aspects <strong>of</strong><br />
switching instructions cannot be carried out.<br />
Some time pressure.<br />
Possibly some difficulties in operating or physically handling <strong>the</strong><br />
equipment.<br />
Yes – proceduralised within equipment types. Occasional non-routine<br />
activities required to cope with unusual/unfamiliar equipment, or<br />
equipment not owned by NGT.<br />
Yes – <strong>of</strong>ten. Except in cases with unfamiliar equipment.<br />
18
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 13 – Phase 4: Report back to NOC<br />
Goal Specification<br />
Cue identification<br />
Expectancy<br />
Conceptual Model<br />
Uncertainty<br />
Information<br />
Situation Awareness<br />
Situation<br />
Assessment<br />
Options<br />
Stress<br />
Choice<br />
Analogy<br />
Inform NOC <strong>of</strong> isolation status.<br />
Switching telephone.<br />
NOC operator answers.<br />
NOC accepts.<br />
Manner in which circuit is now isolated.<br />
Form <strong>of</strong> procedures.<br />
No – possibly fur<strong>the</strong>r instructions, possibly mismatches local situation<br />
and remote displays in NOC.<br />
Switching log.<br />
Verbal information from NOC.<br />
Switching log.<br />
Yes – all information used.<br />
No (raise or add on fur<strong>the</strong>r requests etc. to <strong>the</strong> same call)<br />
No<br />
Yes – highly proceduralised<br />
Yes – frequently performed activity<br />
From <strong>the</strong> CDM outputs, propositional networks were constructed for each incident phase.<br />
The propositional networks are presented in Figure 3-5 to Figure 3-9. Propositional<br />
networks consist <strong>of</strong> a set <strong>of</strong> nodes that represent sources <strong>of</strong> information, agents, and<br />
objects etc. that are linked through specific causal paths. From this network, it is possible<br />
to identify required information and possible options relevant to this incident. The<br />
concept behind using a propositional network in this manner is that it represents <strong>the</strong><br />
‘ideal’ collection <strong>of</strong> knowledge for <strong>the</strong> scenario. As <strong>the</strong> incident unfolds, so participants<br />
will have access to more <strong>of</strong> this knowledge (ei<strong>the</strong>r through communication with o<strong>the</strong>r<br />
agents or through recognising changes in <strong>the</strong> incident status). Consequently, within this<br />
propositional network, Situation Awareness can be represented as <strong>the</strong> change in<br />
weighting <strong>of</strong> links. Propositional networks were developed for <strong>the</strong> overall scenario and<br />
also <strong>the</strong> incident phases identified during <strong>the</strong> CDM analysis. The propositional networks<br />
indicate which <strong>of</strong> <strong>the</strong> knowledge objects are active (i.e. agents are using <strong>the</strong>m) during<br />
each incident phase. The light blue nodes in <strong>the</strong> propositional networks represent<br />
unactivated knowledge objects (i.e. knowledge is available but is not required nor is it<br />
being used). The red nodes represent active (or currently being used) knowledge objects.<br />
19
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Gas Insulated<br />
Circuit<br />
Breakers<br />
Airblast<br />
Notices/Locks<br />
NOC<br />
Check Open<br />
Lock & Caution<br />
Location<br />
SystemState<br />
Certificates<br />
Isolation<br />
Control<br />
Engineer<br />
Open Lock &<br />
Caution<br />
Accept<br />
Instructions<br />
Switching<br />
Wokingham<br />
Shutters<br />
Time<br />
Dressed<br />
Log Sheet<br />
Procedures<br />
Rear<br />
Refuse<br />
Switching Log<br />
Displays<br />
Front<br />
Earth Switches<br />
Isolators<br />
WE1000<br />
Points <strong>of</strong><br />
Isolation<br />
Open<br />
Busbar<br />
Circuits<br />
Electrical<br />
Contacts<br />
Closed<br />
Phone<br />
Current<br />
Voltage<br />
Faulty<br />
DBI Indication<br />
Switching<br />
Phone<br />
Equipment<br />
Lables<br />
Identity<br />
SAP<br />
Substations<br />
Figure 3-5 – Propositional Network for Objects referred to in CDM tables<br />
Gas Insulated<br />
Circuit<br />
Breakers<br />
Airblast<br />
Notices/Locks<br />
NOC<br />
Check Open<br />
Lock & Caution<br />
Location<br />
SystemState<br />
Certificates<br />
Isolation<br />
Control<br />
Engineer<br />
Open Lock &<br />
Caution<br />
Accept<br />
Instructions<br />
Switching<br />
Wokingham<br />
Shutters<br />
Time<br />
Dressed<br />
Log Sheet<br />
Procedures<br />
Rear<br />
Refuse<br />
Switching Log<br />
Displays<br />
Front<br />
Earth Switches<br />
Isolators<br />
WE1000<br />
Points <strong>of</strong><br />
Isolation<br />
Open<br />
Busbar<br />
Circuits<br />
Electrical<br />
Contacts<br />
Closed<br />
Phone<br />
Current<br />
Voltage<br />
Faulty<br />
DBI Indication<br />
Switching<br />
Phone<br />
Equipment<br />
Lables<br />
Identity<br />
SAP<br />
Substations<br />
Figure 3-6 – Propositional Network for CDM Phase 1<br />
20
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Gas Insulated<br />
Circuit<br />
Breakers<br />
Airblast<br />
Notices/Locks<br />
NOC<br />
Check Open<br />
Lock &Caution<br />
Location<br />
SystemState<br />
Certificates<br />
Isolation<br />
Control<br />
Engineer<br />
Open Lock &<br />
Caution<br />
Accept<br />
Instructions<br />
Switching<br />
Wokingham<br />
Shutters<br />
Time<br />
Dressed<br />
Log Sheet<br />
Procedures<br />
Rear<br />
Refuse<br />
Switching Log<br />
Displays<br />
Front<br />
Earth Switches<br />
Isolators<br />
WE1000<br />
Points <strong>of</strong><br />
Isolation<br />
Open<br />
Busbar<br />
Circuits<br />
Electrical<br />
Contacts<br />
Closed<br />
Phone<br />
Current<br />
Voltage<br />
Faulty<br />
DBI Indication<br />
Switching<br />
Phone<br />
Equipment<br />
Lables<br />
Identity<br />
SAP<br />
Substations<br />
Figure 3-7 – Propositional Network for CDM phase two<br />
Gas Insulated<br />
Circuit<br />
Breakers<br />
Airblast<br />
Notices/Locks<br />
NOC<br />
Check Open<br />
Lock &Caution<br />
Location<br />
SystemState<br />
Certificates<br />
Isolation<br />
Control<br />
Engineer<br />
Open Lock &<br />
Caution<br />
Accept<br />
Instructions<br />
Switching<br />
Wokingham<br />
Shutters<br />
Time<br />
Dressed<br />
Log Sheet<br />
Procedures<br />
Rear<br />
Refuse<br />
Switching Log<br />
Displays<br />
Front<br />
Earth Switches<br />
Isolators<br />
WE1000<br />
Points <strong>of</strong><br />
Isolation<br />
Open<br />
Busbar<br />
Circuits<br />
Electrical<br />
Contacts<br />
Closed<br />
Phone<br />
Current<br />
Voltage<br />
Faulty<br />
DBI Indication<br />
Switching<br />
Phone<br />
Equipment<br />
Lables<br />
Identity<br />
SAP<br />
Substations<br />
Figure 3-8 – Propositional Network for CDM phase three<br />
21
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Gas Insulated<br />
Circuit<br />
Breakers<br />
Airblast<br />
Notices/Locks<br />
NOC<br />
Check Open<br />
Lock & Caution<br />
Location<br />
SystemState<br />
Certificates<br />
Isolation<br />
Control<br />
Engineer<br />
Open Lock &<br />
Caution<br />
Accept<br />
Instructions<br />
Switching<br />
Wokingham<br />
Shutters<br />
Time<br />
Dressed<br />
Log Sheet<br />
Procedures<br />
Rear<br />
Refuse<br />
Switching Log<br />
Displays<br />
Front<br />
Earth Switches<br />
Isolators<br />
WE1000<br />
Points <strong>of</strong><br />
Isolation<br />
Open<br />
Busbar<br />
Circuits<br />
Electrical<br />
Contacts<br />
Closed<br />
Phone<br />
Current<br />
Voltage<br />
Faulty<br />
DBI Indication<br />
Switching<br />
Phone<br />
Equipment<br />
Lables<br />
Identity<br />
SAP<br />
Substations<br />
Figure 3-9 – Propositional Network for CDM phase four<br />
The results from <strong>the</strong> EAST analysis indicate that <strong>the</strong> key nodes in <strong>the</strong> Barking scenario<br />
network were <strong>the</strong> NOC operator and <strong>the</strong> SAP at Barking substation. The majority <strong>of</strong> <strong>the</strong><br />
communications observed occurred between <strong>the</strong> NOC and <strong>the</strong> SAP, and <strong>the</strong> operational<br />
loading figures indicate that <strong>the</strong> SAP at Barking conducted <strong>the</strong> majority <strong>of</strong> <strong>the</strong> work<br />
required, on instruction from <strong>the</strong> NOC operator. Over half <strong>of</strong> all <strong>the</strong> tasks involved were<br />
identified as teamwork tasks, and a medium level <strong>of</strong> co-ordination was calculated<br />
between <strong>the</strong> agents involved. The EAST analysis indicated that NOC operator assumes<br />
<strong>the</strong> role <strong>of</strong> commander during <strong>the</strong> scenario, and all work is undertaken only on his<br />
instruction.<br />
Advantages<br />
• The EAST approach <strong>of</strong>fers an exhaustive analysis <strong>of</strong> C4i activity. Activity<br />
is analysed using a number <strong>of</strong> different approaches, ensuring that a<br />
comprehensive analysis <strong>of</strong> <strong>the</strong> scenario under analysis is achieved.<br />
• The EAST output is extremely useful, <strong>of</strong>fering a number <strong>of</strong> different<br />
analyses and perspectives on <strong>the</strong> C4i activity in question. Each component<br />
method <strong>of</strong>fers a different representation or analysis <strong>of</strong> <strong>the</strong> C4i activity<br />
observed. For example, <strong>the</strong> SNA provides an indication <strong>of</strong> <strong>the</strong> key agents<br />
within <strong>the</strong> scenario and identifies communication links, <strong>the</strong> CDA identifies<br />
teamwork tasks and provides a co-ordination rating for each teamwork task<br />
step, and <strong>the</strong> propositional network component identifies <strong>the</strong> ideal<br />
knowledge (i.e. SA) required for effective task performance.<br />
22
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Disadvantages<br />
• The methods within <strong>the</strong> EAST framework are conceptually simple, ensuring<br />
that <strong>the</strong> methodology requires minimal training and is relatively easy to<br />
apply.<br />
• The provision <strong>of</strong> WESTT and Agna s<strong>of</strong>tware packages reduces <strong>the</strong> EAST<br />
application time considerably.<br />
• The EAST methodology is generic, allowing it to be applied in any domain<br />
where C4i activity takes place. To date, <strong>the</strong> EAST methodology has been<br />
used to analyse C4i activity in a number <strong>of</strong> diverse scenarios, including fire<br />
service training scenario’s (Baber et al 2004), military aviation scenarios<br />
(Stewart et al 2004a), railway signalling tasks (Walker et al 2004a), air<br />
traffic control scenarios (Walker et al 2004b), energy distribution scenarios<br />
(Salmon et al 2004) and naval warfare scenarios (Stewart et al 2004b).<br />
• The methodologies that make up EAST share a common <strong>the</strong>oretical<br />
background. This makes integration viable (and embodies a form <strong>of</strong><br />
congruent validity).<br />
• There are possibilities for fur<strong>the</strong>r recombination <strong>of</strong> methods to explore<br />
issues outside <strong>of</strong> <strong>the</strong> EAST framework (Walker et al 2005a).<br />
• All <strong>the</strong> methods have a validation history and prior context <strong>of</strong> use.<br />
• Due to <strong>the</strong> techniques exhaustiveness it is time consuming to apply. The<br />
initial observation, construction <strong>of</strong> <strong>the</strong> HTA and OSD are particularly time<br />
consuming elements <strong>of</strong> <strong>the</strong> methodology.<br />
• The data obtained during <strong>the</strong> CDM analysis a function <strong>of</strong> <strong>the</strong> analysts skill<br />
and <strong>the</strong> quality <strong>of</strong> <strong>the</strong> SME used.<br />
• Reliability <strong>of</strong> <strong>the</strong> technique in some areas is questionable. Much <strong>of</strong> <strong>the</strong><br />
analysis is based upon <strong>the</strong> subjective judgement <strong>of</strong> <strong>the</strong> analyst involved, and<br />
so intra-analyst and inter analyst reliability may suffer.<br />
• For <strong>the</strong> technique to work properly, access to <strong>the</strong> system and personnel<br />
under analysis is required. It is <strong>of</strong>ten difficult to gain sufficient access to <strong>the</strong><br />
system under analysis. Fur<strong>the</strong>r, additional access to SME’s for data<br />
validation purposes is also difficult to obtain.<br />
• The methodology is still evolving and as a consequence, some EAST<br />
outputs may differ in format and presentation.<br />
• Without <strong>the</strong> provision <strong>of</strong> <strong>the</strong> appropriate s<strong>of</strong>tware, an EAST analysis may<br />
become laborious to perform for larger, complex scenarios.<br />
• Some prior knowledge <strong>of</strong> HF methods is required.<br />
23
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Related methods<br />
The EAST methodology is comprised <strong>of</strong> a number <strong>of</strong> HF methods, including HTA,<br />
observation, CDA, CUD, OSD’s, SNA, CDM and propositional networks. The authors<br />
have also developed <strong>the</strong> WESTT methodology, which can be used to automate a large<br />
portion <strong>of</strong> <strong>the</strong> EAST analysis procedure.<br />
Approximate training and application times<br />
The EAST methodology is conceptually simple, however due to <strong>the</strong> number <strong>of</strong><br />
component methods involved, requires a lengthy training session. Some prior experience<br />
in <strong>the</strong> application <strong>of</strong> HF methods is required. For <strong>the</strong> recent applications <strong>of</strong> EAST<br />
described in this report, a training session lasting approximately 8 hours was given to <strong>the</strong><br />
analysts involved. Due to <strong>the</strong> exhaustive nature <strong>of</strong> <strong>the</strong> EAST methodology, and <strong>the</strong><br />
complex nature <strong>of</strong> C4i activity, <strong>the</strong> typical application time is high. The construction <strong>of</strong><br />
<strong>the</strong> HTA and <strong>the</strong> OSD for C4i scenarios is a particularly lengthy process. Conducting an<br />
EAST analysis manually can take up to two weeks for large, complex scenarios. The<br />
application time is significantly reduced by <strong>the</strong> provision <strong>of</strong> <strong>the</strong> WESTT s<strong>of</strong>tware<br />
package, which automates a large portion <strong>of</strong> <strong>the</strong> EAST analysis process.<br />
Reliability and validity<br />
Intra and inter reliability may be questionable in some instances. For example, different<br />
analysts may make different interpretations <strong>of</strong> <strong>the</strong> OSD symbols in <strong>the</strong> OSD glossary.<br />
The use <strong>of</strong> multiple methods within <strong>the</strong> EAST framework heightens <strong>the</strong> potential for<br />
unreliability. It is difficult to assess <strong>the</strong> validity <strong>of</strong> <strong>the</strong> data obtained during an EAST<br />
analysis. Typically, SMEs are used during data collection, and so a high level <strong>of</strong> validity<br />
is assumed.<br />
Congruent validity is favourable due to effective methods integration. Construct validity<br />
also good, as component methods relate well to <strong>the</strong> core descriptive constructs <strong>of</strong> C4i<br />
(Walker et al 2005b). Construct and face validity are also good because many <strong>of</strong> <strong>the</strong><br />
features <strong>of</strong> interest are also observable and manifest in <strong>the</strong> environment <strong>of</strong> interest. The<br />
CDM complements this approach by tapping into <strong>the</strong> ‘implicit’ parts <strong>of</strong> <strong>the</strong> decision<br />
making process.<br />
Tools needed<br />
Due to <strong>the</strong> various component analyses involved in an EAST analysis, a number <strong>of</strong><br />
different tools and s<strong>of</strong>tware packages are required. For <strong>the</strong> observation and CDM phases<br />
<strong>of</strong> an EAST analysis, video and audio recording equipment is required. A word<br />
processing s<strong>of</strong>tware package such as Micros<strong>of</strong>t Word is required to transcribe <strong>the</strong><br />
observation and CDM data. The CDA is typically conducted using Micros<strong>of</strong>t Excel. A<br />
large part (e.g. OSD and SNA) <strong>of</strong> <strong>the</strong> EAST procedure can be automated by using <strong>the</strong><br />
WESTT s<strong>of</strong>tware package. Alternatively, <strong>the</strong> OSD CUD, and propositional networks are<br />
typically drawn using Micros<strong>of</strong>t Visio. The SNA is conducted using <strong>the</strong> Agna SNA<br />
s<strong>of</strong>tware package.<br />
24
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Flowchart<br />
START<br />
Define scenarios<br />
Conduct HTA for<br />
scenario under<br />
analysis<br />
Plan observation<br />
Conduct observation<br />
and CDM analysis<br />
Create scenario<br />
transcript<br />
Re-iterate HTA<br />
Conduct CDA<br />
Construct CUD<br />
Conduct SNA<br />
Construct OSD<br />
Construct Prop<br />
networks<br />
STOP<br />
25
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
3.2 Observation<br />
Background and applications<br />
Observational techniques are used to ga<strong>the</strong>r data regarding <strong>the</strong> physical and verbal<br />
aspects <strong>of</strong> a particular task or scenario. Observational techniques are used to collect data<br />
regarding various aspects <strong>of</strong> system and task performance, such as data regarding <strong>the</strong><br />
tasks catered for by <strong>the</strong> system, <strong>the</strong> individuals performing <strong>the</strong> tasks, <strong>the</strong> tasks <strong>the</strong>mselves<br />
(task steps and sequence), errors made, communications between individuals, <strong>the</strong><br />
technology used by <strong>the</strong> system in conducting <strong>the</strong> tasks (controls, displays, communication<br />
technology etc), <strong>the</strong> system environment and <strong>the</strong> organisational environment.<br />
Observation has been extensively used in <strong>the</strong> HF community, and typically forms <strong>the</strong><br />
beginning <strong>of</strong> an analysis effort.<br />
Domain <strong>of</strong> application<br />
Generic.<br />
Procedure and advice<br />
Step 1: Define <strong>the</strong> objective <strong>of</strong> <strong>the</strong> analysis<br />
The first step in observational analysis involves defining <strong>the</strong> aims and objectives <strong>of</strong> <strong>the</strong><br />
observation. This should include determining which product or system is under analysis,<br />
which environment <strong>the</strong> observation will take place, which user groups will be observed,<br />
what type <strong>of</strong> scenario’s will be observed and what data is required. Each point should be<br />
clearly defined and stated before <strong>the</strong> process continues.<br />
Step 2: Define <strong>the</strong> scenario(s)<br />
Once <strong>the</strong> aims and objectives <strong>of</strong> <strong>the</strong> analysis are clearly defined, <strong>the</strong> scenario(s) to be<br />
observed should be defined and described fur<strong>the</strong>r. For example, when conducting an<br />
observational analysis <strong>of</strong> control room operation, which type <strong>of</strong> scenario is required<br />
should be clearly defined. Is normal operation under scrutiny or is <strong>the</strong> analysis focussed<br />
upon operator interaction and performance under emergency situations. The exact nature<br />
<strong>of</strong> <strong>the</strong> required scenario(s) should be clearly defined by <strong>the</strong> observation team. It is<br />
recommended that a HTA is conducted for <strong>the</strong> task or scenario under analysis.<br />
Step 3: Observation plan<br />
Once <strong>the</strong> aim <strong>of</strong> <strong>the</strong> analysis is defined and also <strong>the</strong> type <strong>of</strong> scenario to be observed is<br />
determined, <strong>the</strong> analysis team should proceed to plan <strong>the</strong> observation. The team should<br />
consider what <strong>the</strong>y are hoping to observe, what <strong>the</strong>y are observing, and how <strong>the</strong>y are<br />
going to observe it. Depending upon <strong>the</strong> nature <strong>of</strong> <strong>the</strong> observation, access to <strong>the</strong> system<br />
in question should be gained first. This may involve holding meetings with <strong>the</strong><br />
establishment in question, and is typically a lengthy process. Any recording tools should<br />
be defined and also <strong>the</strong> length <strong>of</strong> observations should be determined. Placement <strong>of</strong> video<br />
and audio recording equipment should also be considered. To make things easier, a<br />
walkthrough <strong>of</strong> <strong>the</strong> system/environment/scenario under analysis is recommended. This<br />
26
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
allows <strong>the</strong> analyst(s) to become familiar with <strong>the</strong> task in terms <strong>of</strong> time taken, location and<br />
also <strong>the</strong> system under analysis.<br />
Step 4: Pilot observation<br />
In any observational study a pilot or practice observation is crucial. This allows <strong>the</strong><br />
analysis team to assess any problems with <strong>the</strong> data collection, such as noise interference<br />
or problems with <strong>the</strong> recording equipment. The quality <strong>of</strong> data collected can also be<br />
tested and also any effects <strong>of</strong> <strong>the</strong> observation upon task performance can be assessed. If<br />
major problems are encountered, <strong>the</strong> observation may have to be re-designed. Steps 1 to 4<br />
should be repeated until <strong>the</strong> analysis team are happy that <strong>the</strong> quality <strong>of</strong> <strong>the</strong> data collected<br />
will be sufficient for <strong>the</strong>ir study requirements.<br />
Step 5: Conduct Observation<br />
Once <strong>the</strong> observation has been designed, <strong>the</strong> team should proceed with <strong>the</strong> observation(s).<br />
Observation length and timing are dependent upon <strong>the</strong> scope and requirements <strong>of</strong> <strong>the</strong><br />
analysis. Once <strong>the</strong> required data is collected, <strong>the</strong> observation should stop and step 6<br />
should be undertaken.<br />
Step 6: Create observation transcript<br />
Once <strong>the</strong> observation is complete, <strong>the</strong> analyst(s) should construct an observation<br />
transcript. An extract <strong>of</strong> an observation transcript from an observation <strong>of</strong> an energy<br />
distribution switching scenario (Salmon et al 2004b) is presented in Table 14. The<br />
observational transcript should contain a description <strong>of</strong> all activity observed during <strong>the</strong><br />
analysis, including <strong>the</strong> agents involved, <strong>the</strong> time, <strong>the</strong> technology used and also any<br />
additional notes (e.g. purpose <strong>of</strong> activity).<br />
Step 7: Data analysis<br />
Once <strong>the</strong> observation is complete, and <strong>the</strong> observation transcripts have been constructed,<br />
<strong>the</strong> data analysis procedure can begin. Depending upon <strong>the</strong> analysis requirements, <strong>the</strong><br />
team should analyse <strong>the</strong> data in <strong>the</strong> format that is required, such as frequency <strong>of</strong> tasks,<br />
verbal interactions, sequence <strong>of</strong> tasks etc. When analysing visual data, typically user<br />
behaviours are coded into specific groups. The s<strong>of</strong>tware package Observer is used to<br />
aid <strong>the</strong> analyst in this process.<br />
27
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 14 – Observation transcript extract (Source Salmon et al 2004b)<br />
Time Process Comms Location Notes<br />
09:54 GA & DW engage in general pre-amble about <strong>the</strong><br />
forthcoming switching ops. Considering asking <strong>the</strong><br />
Operations Centre in Wokingham to switch out SGT1A1B<br />
early so that SGT5 can be done at <strong>the</strong> same time<br />
10:15 Wokingham call Barking ask is <strong>the</strong>y still want isolation –<br />
GA confirms yes.<br />
Person<br />
person<br />
Telephone<br />
to<br />
Barking<br />
275Kv<br />
switchhouse<br />
10:19 Switching phone rings throughout building. NOC Wants<br />
132Kv busbar opened [GA].<br />
Green<br />
Telephone<br />
10:40 SGT1A1B waiting to be handed over to NOC. Delay.<br />
10:40 Wokingham report to Barking 275 [GA] complication with<br />
EDF. EDF want to reselect circuits at <strong>the</strong> last minute at<br />
ano<strong>the</strong>r substation due to planned shutdown <strong>of</strong> SGT1A<br />
10:53 GA and DW discuss EDF problem. Can DW reconfigure<br />
circuits in Barking West (33Kv)<br />
Telephone Wokingham contact EDF to confirm switching (as <strong>the</strong>y did with<br />
Barking 275 at 10:15). EDF report a problem. Wokingham pass<br />
this onto Barking 275<br />
Person to<br />
person<br />
10:53 GA contact Wokingham. Confusion as to what circuits need<br />
reconfiguring and who can do it. GA talks to DW at <strong>the</strong><br />
same time. Decided that DW can reconfigure circuits.<br />
Wokingham give GA name and phone number <strong>of</strong> EDF<br />
contact<br />
10:58 GA and DW discuss plans for Barking West 33. Discuss<br />
who owns what, safety rules etc. Also discuss and decide<br />
order <strong>of</strong> subsequent site visits (might have to go to Barking<br />
West 33 twice)<br />
11:04 GA & DW Waiting to travel to Barking West<br />
GA/DW<br />
Person to<br />
Person,<br />
Wokingham<br />
Telephone<br />
This is an unplanned measure – now need to go to Barking West<br />
33Kv to reconfigure local electricity supply circuits<br />
28
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Step 8: Fur<strong>the</strong>r analysis<br />
Once <strong>the</strong> initial process <strong>of</strong> transcribing and coding <strong>the</strong> observational data is complete,<br />
fur<strong>the</strong>r analysis <strong>of</strong> <strong>the</strong> data begins. Depending upon <strong>the</strong> nature <strong>of</strong> <strong>the</strong> analysis,<br />
observation data is used to inform a number <strong>of</strong> different HF analyses, such as task<br />
analysis, error analysis and communications analysis. Typically, observational data is<br />
used to develop a task analysis (e.g. HTA) <strong>of</strong> <strong>the</strong> task or scenario under analysis.<br />
Step 9: Participant feedback<br />
Once <strong>the</strong> data has been analysed and conclusions have been drawn, <strong>the</strong> participants<br />
involved should be provided with feedback <strong>of</strong> some sort. This could be in <strong>the</strong> form <strong>of</strong> a<br />
feedback session or a letter to each participant. The type <strong>of</strong> feedback used is determined<br />
by <strong>the</strong> analysis team.<br />
Example<br />
An observational analysis <strong>of</strong> a fire service training scenario was conducted as part <strong>of</strong> an<br />
analysis <strong>of</strong> existing C4i activity in civil domains (Baber et al 2004). Three observers<br />
observed <strong>the</strong> fire training service-training scenario, “Hazardous chemical spillage at<br />
remote farmhouse”. The three observers sat in on <strong>the</strong> exercise and recorded <strong>the</strong><br />
discussion <strong>of</strong> <strong>the</strong> participants. The notes from <strong>the</strong> discussion were <strong>the</strong>n collated into a<br />
combined timeline <strong>of</strong> <strong>the</strong> incident. This timeline, and <strong>the</strong> notes taken during <strong>the</strong> exercise,<br />
<strong>the</strong>n formed <strong>the</strong> basis for <strong>the</strong> following HF analyses.<br />
Hierarchical task analysis<br />
Operator sequence diagram<br />
Social network analysis<br />
Co-ordination demand analysis<br />
Comms usage diagram<br />
Critical decision method<br />
The exercise involved a combination <strong>of</strong> focus group discussion with paired activity in<br />
order to define appropriate courses <strong>of</strong> action to deal with <strong>the</strong> specified incident. The class<br />
facilitator provided <strong>the</strong> initial description <strong>of</strong> an incident, i.e. a report <strong>of</strong> possible<br />
hazardous materials on a remote farm, and <strong>the</strong>n added additional information as <strong>the</strong><br />
incident unfolded, e.g. reports <strong>of</strong> casualties, problems with labelling on hazardous<br />
materials etc. The exercise was designed to encourage experienced fire-fighters to<br />
consider risks arising from hazardous materials and <strong>the</strong> appropriate courses <strong>of</strong> action <strong>the</strong>y<br />
would need to take, e.g. in terms <strong>of</strong> protective equipment, incident management,<br />
information seeking activities etc. From <strong>the</strong> data obtained during <strong>the</strong> observation, an<br />
event flowchart was constructed, which acted as <strong>the</strong> primary input to <strong>the</strong> analysis<br />
techniques used. The event flowchart is presented in Figure 3-10.<br />
29
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Police called to break-in<br />
at remote farmhouse<br />
Control seeks description<br />
and informs police<br />
<strong>of</strong>ficer<br />
Police <strong>of</strong>ficer proceeds<br />
to farmhouse<br />
Child admitted to<br />
hospital with respiratory<br />
problems<br />
Control inform Police<br />
<strong>of</strong>ficer hazardous<br />
materials on farm<br />
Hospital inform police <strong>of</strong><br />
hazardous materials at<br />
farm<br />
Police control call fire<br />
brigade<br />
Police <strong>of</strong>ficer sets up<br />
outer cordon at farm<br />
Fire squadron proceed to<br />
farmhouse<br />
Fire brigade set up inner<br />
cordon at farm<br />
Hospital contact fire<br />
control requesting urgent<br />
diagnosis <strong>of</strong> chemical<br />
substance<br />
Fire brigade discuss<br />
protection options<br />
Fire brigade conduct<br />
search activity<br />
Locate chemical drums<br />
Chemical diagnosis<br />
Chemical diagnosis questioned as<br />
substance found is powder and not<br />
liquid<br />
Check UN number on<br />
chem.database<br />
Check UN number on<br />
Chemsafe database<br />
Contact supplier<br />
Identify substance<br />
Control inform hospital<br />
<strong>of</strong> diagnosis<br />
Begin decontamination<br />
process<br />
Figure 3-10 - Hazardous chemical spillage event flowchart<br />
30
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Flowchart<br />
START<br />
Define study requirements<br />
Define scenario(s) to be observed<br />
Prepare/design observation<br />
Conduct pilot observation session<br />
Are <strong>the</strong>re any<br />
problems<br />
Y<br />
Conduct observation <strong>of</strong> scenario(s)<br />
N<br />
For data analysis, choose from <strong>the</strong> following based on study/data<br />
requirements:<br />
• Transcribe scenario<br />
• Record task sequences<br />
• Record task times<br />
• Record any errors observed<br />
• Record frequency <strong>of</strong> tasks<br />
STOP<br />
31
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Advantages<br />
Disadvantages<br />
• Observation technique data provides a ‘real life’ insight into man-machine,<br />
and team interaction.<br />
• Various data can be elicited from an observational study, including task<br />
sequences, task analysis, error data, task times, verbal interaction and task<br />
performance.<br />
• Observation has been used extensively in a wide range <strong>of</strong> domains.<br />
• Observation provides objective information.<br />
• Detailed physical task performance data is recorded, including social<br />
interactions and any environmental task influences (Kirwan & Ainsworth<br />
1992).<br />
• Observation is excellent for <strong>the</strong> initial stages <strong>of</strong> <strong>the</strong> task analysis procedure.<br />
• Observation analysis can be used to highlight problems with existing<br />
operational systems. It can be used in this way to inform <strong>the</strong> design <strong>of</strong> new<br />
systems or devices.<br />
• Specific Scenarios are observed in <strong>the</strong>ir ‘real world’ setting.<br />
• Observational data is used as <strong>the</strong> core input into numerous HF analyses<br />
techniques, such as human error identification techniques (SHERPA), task<br />
analysis (HTA), communications analysis (Comms Usage Diagrams), and<br />
charting techniques (operator sequence diagrams).<br />
• Observational techniques are intrusive to task performance. Knowing that<br />
<strong>the</strong>y are being watched tends to elicit new and different behaviours in<br />
participants. For example, when observing control room operators, <strong>the</strong>y<br />
may perform exactly as <strong>the</strong>ir procedures say <strong>the</strong>y should. However, when<br />
not being observed, <strong>the</strong> same control room operator’s may perform<br />
completely differently, using short-cuts and behaviours that are not stated in<br />
<strong>the</strong>ir procedures. This may be due to <strong>the</strong> fact that <strong>the</strong> operator’s do not<br />
wish to be caught bending <strong>the</strong> rules in any way i.e. bypassing a certain<br />
procedure. This effect on behaviour can bias <strong>the</strong> observational data that is<br />
obtained.<br />
• Observational techniques are time consuming in <strong>the</strong>ir application,<br />
particularly <strong>the</strong> data analysis procedure. Kirwan & Ainsworth (1992)<br />
suggest that when conducting <strong>the</strong> transcription process, 1 hour <strong>of</strong> recorded<br />
audio data takes on analyst approximately 8 hours to transcribe.<br />
32
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Related methods<br />
• Cognitive aspects <strong>of</strong> <strong>the</strong> task under analysis are not elicited using<br />
observational techniques. Verbal protocol analysis is more suited for<br />
collecting data on <strong>the</strong> cognitive aspects <strong>of</strong> task performance.<br />
• An observational study can be both difficult and expensive to set up and<br />
conduct. Gaining access to <strong>the</strong> required establishment is <strong>of</strong>ten extremely<br />
difficult and very time consuming. Observational techniques are also costly,<br />
as <strong>the</strong>y require <strong>the</strong> use <strong>of</strong> expensive recording equipment (digital video<br />
camera, audio recording devices).<br />
• Causality is a problem. Errors can be observed and recorded during an<br />
observation but why <strong>the</strong> errors occur may not always be clear.<br />
• The analyst has a very low level <strong>of</strong> experimental control.<br />
• In most cases, a team <strong>of</strong> analysts is required to perform an observation<br />
study. It is <strong>of</strong>ten difficult to acquire a suitable team with sufficient<br />
experience in conducting observational studies.<br />
There are a number <strong>of</strong> different observational techniques, including indirect observation,<br />
participant observation and remote observation. O<strong>the</strong>r related techniques include verbal<br />
protocol analysis, critical decision method, applied cognitive task analysis, walkthroughs<br />
and cognitive walkthroughs. All <strong>of</strong> <strong>the</strong>se techniques require some sort <strong>of</strong> task<br />
observation. Observational data is also used as an input to numerous HF techniques, such<br />
as task analysis, human error identification, and charting techniques.<br />
Approximate training and application times<br />
Whilst <strong>the</strong> training time for an observational analysis is low (Stanton & Young 1999), <strong>the</strong><br />
application time is normally high. The data analysis stage can be particularly time<br />
consuming. Kirwan & Ainsworth (1992) suggest that in <strong>the</strong> transcription process, 1 hour<br />
<strong>of</strong> audio recorded data would take approximately 8 hours to transcribe.<br />
Reliability and validity<br />
Observational analysis is beset by a number <strong>of</strong> problems that can potentially affect <strong>the</strong><br />
reliability and validity <strong>of</strong> <strong>the</strong> technique. According to Baber & Stanton (1996) problems<br />
with causality, bias (in a number <strong>of</strong> forms), construct validity, external validity and<br />
internal validity can all arise unless <strong>the</strong> correct precautions are taken. Whilst<br />
observational techniques possess a high level <strong>of</strong> face validity (Drury 1990) and ecological<br />
validity (Baber & Stanton 1996), analyst or participant bias can adversely affect <strong>the</strong><br />
reliability and validity <strong>of</strong> <strong>the</strong> techniques.<br />
Tools needed<br />
For a thorough observational analysis, <strong>the</strong> appropriate visual and audio recording<br />
equipment is necessary. Observational studies can be conducted using pen and paper,<br />
33
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
however this is not recommended, as crucial parts <strong>of</strong> data are <strong>of</strong>ten not recorded. For <strong>the</strong><br />
data analysis process, a PC with <strong>the</strong> Observer s<strong>of</strong>tware is required.<br />
3.3 HTA – Hierarchical Task <strong>Analysis</strong><br />
Background and Applications<br />
Hierarchical task analysis (HTA) (Annett 2004) is <strong>the</strong> most popular task analysis<br />
technique and has become perhaps <strong>the</strong> most widely used <strong>of</strong> all HF techniques. HTA<br />
involves describing <strong>the</strong> task under analysis through <strong>the</strong> break down <strong>of</strong> <strong>the</strong> task into a<br />
hierarchy <strong>of</strong> goals, sub goals, operations and plans. Goals are <strong>the</strong> unobservable task goals<br />
associated with <strong>the</strong> task in question, operations are <strong>the</strong> observable behaviours required to<br />
accomplish <strong>the</strong> task goals and plans are <strong>the</strong> unobservable decisions and planning made on<br />
behalf <strong>of</strong> <strong>the</strong> operator. The end result is an exhaustive description <strong>of</strong> task activity. HTA<br />
has been applied across a wide spectrum <strong>of</strong> domains, including <strong>the</strong> process control and<br />
power generation industries (Annett 2004), emergency services (Baber et al 2004)<br />
military applications (Kirwan & Ainsworth, 1992; Ainsworth & Marshall, 1998/2000),<br />
civil aviation (Marshall et al 2003), driving, public technology (Stanton and Stevenage<br />
1998) and even retail (shepherd 2001). One <strong>of</strong> <strong>the</strong> main reasons behind <strong>the</strong> enduring<br />
popularity <strong>of</strong> HTA is its flexibility and <strong>the</strong> scope for fur<strong>the</strong>r analysis <strong>of</strong> <strong>the</strong> sub-goal<br />
hierarchy that it <strong>of</strong>fers to <strong>the</strong> HF practitioner. The majority <strong>of</strong> HF analysis methods ei<strong>the</strong>r<br />
require an initial HTA <strong>of</strong> <strong>the</strong> task(s) under analysis as <strong>the</strong>ir input, or at least are made<br />
significantly easier through <strong>the</strong> provision <strong>of</strong> a HTA. HTA acts as an input into numerous<br />
HF analyses techniques, such human error identification (HEI), allocation <strong>of</strong> function,<br />
workload assessment, interface design and evaluation and many more. In a review <strong>of</strong><br />
ergonomics texts, Stanton (2004) highlights at least twelve additional applications to<br />
which HTA has been put, including interface design and evaluation, training, allocation<br />
<strong>of</strong> functions, job description, work organisation, manual design, job aid design, error<br />
prediction and analysis, team task analysis, workload assessment and procedure design.<br />
Domain <strong>of</strong> application<br />
HTA was originally developed for <strong>the</strong> chemical processing and power generation<br />
industries (Annett 2004). However <strong>the</strong> technique is generic and can be applied in any<br />
domain.<br />
Procedure and advice<br />
Step 1: Define task under analysis<br />
The first step in conducting a HTA is to clearly define <strong>the</strong> task(s) under analysis. As well<br />
as identifying <strong>the</strong> task under analysis, <strong>the</strong> purpose <strong>of</strong> <strong>the</strong> task analysis effort should also<br />
be defined. For example, Marshall et al (2003) conducted a HTA <strong>of</strong> a civil aircraft<br />
landing task in order to predict design induced error for <strong>the</strong> flight task in question.<br />
Step 2: Data collection process<br />
Once <strong>the</strong> task under analysis is clearly defined, specific data regarding <strong>the</strong> task should be<br />
collected. The data collected during this process is used to inform <strong>the</strong> development <strong>of</strong> <strong>the</strong><br />
34
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
HTA. Data regarding <strong>the</strong> task steps involved, <strong>the</strong> technology used, interaction between<br />
man and machine and team members, decision making and task constraints should be<br />
collected. There are a number <strong>of</strong> ways to collect this data, including observations,<br />
interviews, and questionnaires. The technique used is dependent upon <strong>the</strong> analysis effort<br />
and <strong>the</strong> various constraints imposed, such as time and access constraints. Once sufficient<br />
data regarding <strong>the</strong> task under analysis is collected, <strong>the</strong> development <strong>of</strong> <strong>the</strong> HTA should<br />
begin.<br />
Step 3: Determine <strong>the</strong> overall goal <strong>of</strong> <strong>the</strong> task<br />
The overall task goal <strong>of</strong> <strong>the</strong> task under analysis should first be specified at <strong>the</strong> top <strong>of</strong> <strong>the</strong><br />
hierarchy i.e. ‘Land aircraft X at New Orleans Airport using <strong>the</strong> Auto-land system’<br />
(Marshall et al 2003), ‘Boil kettle’, or ‘Listen to in-car entertainment’ (Stanton and<br />
Young 1999).<br />
Step 4: Determine task sub-goals<br />
Once <strong>the</strong> overall task goal has been specified, <strong>the</strong> next step is to break this overall goal<br />
down into meaningful sub-goals (usually four or five but this is not rigid), which toge<strong>the</strong>r<br />
form <strong>the</strong> tasks required to achieve <strong>the</strong> overall goal. In <strong>the</strong> task, ‘Land aircraft X at New<br />
Orleans Airport using <strong>the</strong> Auto-land system’ (Marshall et al 2003), <strong>the</strong> overall goal <strong>of</strong><br />
landing <strong>the</strong> aircraft was broken down into <strong>the</strong> sub-goals, ‘Set up for approach’, ‘Line up<br />
aircraft for runway’ and ‘Prepare aircraft for landing’. In a HTA <strong>of</strong> a Ford in-car radio<br />
(Stanton & Young 1999) <strong>the</strong> overall task goal, “Listen to in car entertainment”, was<br />
broken down into <strong>the</strong> following sub-goals, ‘Check unit status’, ‘Press on/<strong>of</strong>f button’,<br />
‘Listen to <strong>the</strong> radio’, ‘Listen to cassette’, and ‘Adjust audio preferences’.<br />
Step 5: Sub-goal decomposition<br />
Next, <strong>the</strong> analyst should break down <strong>the</strong> sub-goals identified in step four into fur<strong>the</strong>r subgoals<br />
and operations, according to <strong>the</strong> task step in question. This process should go on<br />
until an appropriate operation is reached. The bottom level <strong>of</strong> any branch in a HTA will<br />
always be an operation. Whilst everything above an operation specifies goals, operations<br />
actually say what needs to be done. Thus operations are actions to be made by <strong>the</strong><br />
operator in order to achieve <strong>the</strong> associated goal. For example, in <strong>the</strong> HTA <strong>of</strong> <strong>the</strong> flight<br />
task ‘Land aircraft X at New Orleans Airport using <strong>the</strong> Auto-land system’ (Marshall et al<br />
2003), <strong>the</strong> sub-goal ‘Reduce airspeed to 210 Knots’ is broken down into <strong>the</strong> following<br />
operations, ‘Check current airspeed’ and ‘Dial <strong>the</strong> Speed/MACH selector knob to enter<br />
210 on <strong>the</strong> IAS/MACH display’.<br />
Step 6: Plans analysis<br />
Once all <strong>of</strong> <strong>the</strong> sub-goals and operations have been fully described, <strong>the</strong> plans need to be<br />
added. Plans dictate how <strong>the</strong> goals are achieved. A simple plan would say Do 1, <strong>the</strong>n 2,<br />
and <strong>the</strong>n 3. Once <strong>the</strong> plan is completed, <strong>the</strong> operator returns to <strong>the</strong> super-ordinate level.<br />
Plans do not have to be linear and can come in any form such as Do 1, Or 2 and 3. The<br />
different types <strong>of</strong> plans used are presented in Table 15.<br />
35
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 15 - Example HTA plans<br />
Plan<br />
Example<br />
Linear Do 1 <strong>the</strong>n 2 <strong>the</strong>n 3<br />
Non-linear<br />
Do 1, 2 and 3 in any order<br />
Simultaneous<br />
Do 1, <strong>the</strong>n 2 and 3 at <strong>the</strong> same time<br />
Branching<br />
Do 1, if X present <strong>the</strong>n do 2 <strong>the</strong>n 3, if X is not<br />
present <strong>the</strong>n EXIT<br />
Cyclical<br />
Selection Do 1 <strong>the</strong>n 2 or 3<br />
Do 1 <strong>the</strong>n 2 <strong>the</strong>n 3 and repeat until X<br />
Advantages<br />
Disadvantages<br />
• HTA is a technique that is both easy to learn and easy to implement.<br />
• The output <strong>of</strong> a HTA is extremely useful and forms <strong>the</strong> input for numerous<br />
HF analyses, such as error analysis, interface design and evaluation and<br />
allocation <strong>of</strong> function analysis.<br />
• Quick to use in most instances.<br />
• Comprehensive technique covers all sub-tasks <strong>of</strong> <strong>the</strong> task in question.<br />
• HTA has been used extensively in a wide range <strong>of</strong> contexts.<br />
• Conducting an HTA gives <strong>the</strong> user a great insight into <strong>the</strong> task under<br />
analysis.<br />
• HTA is an excellent technique to use when requiring a task description for<br />
fur<strong>the</strong>r analysis. If performed correctly, <strong>the</strong> HTA should depict everything<br />
that needs to be done in order to complete <strong>the</strong> task in question.<br />
• The technique is generic and can be applied to any task in any domain.<br />
• Tasks can be analysed to any required level <strong>of</strong> detail, depending on <strong>the</strong><br />
purpose.<br />
• Provides mainly descriptive information ra<strong>the</strong>r than analytical information.<br />
• HTA contains little that can be used directly to provide design solutions.<br />
• HTA does not cater for <strong>the</strong> cognitive components <strong>of</strong> <strong>the</strong> task under analysis.<br />
36
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Related Methods<br />
• The technique may become laborious and time consuming to conduct for<br />
more complex and larger tasks.<br />
• The initial data collection phase is time consuming and requires <strong>the</strong> analyst<br />
to be competent in a variety <strong>of</strong> HF techniques, such as interviews,<br />
observations and questionnaires.<br />
• The reliability <strong>of</strong> <strong>the</strong> technique may be questionable in some instances. For<br />
example, for <strong>the</strong> same task, different analysts may produce very different<br />
sub-goal hierarchies.<br />
• HTA has <strong>of</strong>ten been labelled as an art ra<strong>the</strong>r than a science.<br />
• An adequate s<strong>of</strong>tware version <strong>of</strong> HTA has yet to emerge.<br />
HTA is widely used in HF and <strong>of</strong>ten forms <strong>the</strong> first step in a number <strong>of</strong> analyses, such as<br />
HEI, HRA and mental workload assessment. In a review <strong>of</strong> ergonomics texts, Stanton<br />
(2004) highlights at least twelve additional applications to which HTA has been put,<br />
including interface design and evaluation, training, allocation <strong>of</strong> functions, job<br />
description, work organisation, manual design, job aid design, error prediction and<br />
analysis, team task analysis, workload assessment and procedure design. As a result<br />
HTA is perhaps <strong>the</strong> most commonly used HF method, and is typically used as <strong>the</strong> start<br />
point or basis <strong>of</strong> any HF analysis.<br />
Approximate Training and Application Times<br />
The training time for <strong>the</strong> HTA method is largely dependent upon on <strong>the</strong> skill <strong>of</strong> <strong>the</strong><br />
analysts involved. It is <strong>of</strong>ten said that using HTA is more <strong>of</strong> an art than a science, and <strong>the</strong><br />
training time required reflects this. Some analysts may pick up <strong>the</strong> method immediately,<br />
whilst o<strong>the</strong>rs may use <strong>the</strong> technique repeatedly without becoming skilled in its use. It is<br />
<strong>the</strong>refore estimated that <strong>the</strong> training time for HTA is high. A survey by Ainsworth &<br />
Marshall (1998/2000) found that <strong>the</strong> more experienced practitioners produced more<br />
complete and acceptable analyses. Stanton & Young (1999) report that <strong>the</strong> training and<br />
application time for HTA is substantial. The application time associated with HTA is<br />
dependent upon <strong>the</strong> size and complexity <strong>of</strong> <strong>the</strong> task under analysis. For large, complex<br />
tasks, <strong>the</strong> application time for HTA would be high.<br />
Reliability and Validity<br />
According to Annett (2004), <strong>the</strong> reliability and validity <strong>of</strong> HTA is not easily assessed.<br />
From a comparison <strong>of</strong> twelve HF techniques, Stanton & Young (1999) reported that, <strong>the</strong><br />
technique achieved an acceptable level <strong>of</strong> validity but a poor level <strong>of</strong> reliability. The<br />
reliability <strong>of</strong> <strong>the</strong> technique is certainly questionable. It seems that different analysts, with<br />
different experience may produce entirely different analyses for <strong>the</strong> same task (intraanalyst<br />
reliability). Similarly, <strong>the</strong> same analyst may produce different analyses on<br />
different occasions for <strong>the</strong> same task (inter-analyst reliability).<br />
37
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Tools needed.<br />
HTA can be carried out using only pencil and paper. The HTA output can be developed<br />
and presented in a number <strong>of</strong> s<strong>of</strong>tware applications, such as Micros<strong>of</strong>t Visio, Micros<strong>of</strong>t<br />
Word and Micros<strong>of</strong>t Excel. A number <strong>of</strong> HTA s<strong>of</strong>tware tools also exist, such as <strong>the</strong> HFIT<br />
DTC’s HTA tool.<br />
Flowchart<br />
START<br />
State overall goal<br />
State subordinate operations<br />
Select next operation<br />
State plan<br />
Check <strong>the</strong> adequacy <strong>of</strong><br />
rediscription<br />
Revise<br />
rediscription<br />
Is redescription<br />
ok<br />
Consider <strong>the</strong> first/next<br />
suboperation<br />
Is fur<strong>the</strong>r<br />
redescription<br />
required<br />
Terminate <strong>the</strong> redescription <strong>of</strong> this<br />
operation<br />
Are <strong>the</strong>re any<br />
more<br />
operations<br />
STOP<br />
38
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
3.4 CDA – Co-Ordination Demands <strong>Analysis</strong><br />
Background and application<br />
Co-ordination demands analysis (CDA) is used to rate <strong>the</strong> co-ordination between agents<br />
involved in teamwork activity. CDA is used to identify <strong>the</strong> extent to which team<br />
members have to work with each o<strong>the</strong>r in order to accomplish <strong>the</strong> task(s) under analysis.<br />
The CDA procedure allows analysts to identify <strong>the</strong> operational skills required within<br />
team tasks, but also <strong>the</strong> teamwork skills needed for smooth coordination among team<br />
members. In its present usage teamwork skills are extracted from a HTA (Annett 2004)<br />
and rated on a scale <strong>of</strong> 1 (low) to 3 (high) against each behaviour in <strong>the</strong> co-ordination<br />
behaviour taxonomy presented in Table 16 (Source: Burke 2004). From <strong>the</strong>se individual<br />
ratings a ‘total coordination’ figure for each teamwork task step is derived.<br />
Table 16 - CDA teamwork taxonomy (Source: Burke 2004)<br />
Coordination<br />
Dimension<br />
Communication<br />
Situational<br />
Awareness (SA)<br />
Decision Making<br />
(DM)<br />
Mission analysis<br />
(MA)<br />
Leadership<br />
Adaptability<br />
Assertiveness<br />
Total Coordination<br />
Definition<br />
Includes sending, receiving, and acknowledging information among<br />
crewmembers.<br />
Refers to identifying <strong>the</strong> source and nature <strong>of</strong> problems, maintaining an<br />
accurate perception <strong>of</strong> <strong>the</strong> aircraft’s location relative to <strong>the</strong> external<br />
environment, and detecting situations that require action<br />
Includes identifying possible solutions to problems, evaluating <strong>the</strong><br />
consequences <strong>of</strong> each alternative, selecting <strong>the</strong> best alternative, and<br />
ga<strong>the</strong>ring information needed prior to arriving at a decision<br />
Includes monitoring, allocating, and coordinating <strong>the</strong> resources <strong>of</strong> <strong>the</strong> crew<br />
and aircraft; prioritizing tasks; setting goals and developing plans to<br />
accomplish <strong>the</strong> goals; creating contingency plans.<br />
Refers to directing activities <strong>of</strong> o<strong>the</strong>rs, monitoring and assessing <strong>the</strong><br />
performance <strong>of</strong> crew members, motivating members, and communicating<br />
mission requirements<br />
Refers to <strong>the</strong> ability to alter one’s course <strong>of</strong> action as necessary, maintain<br />
constructive behaviour under pressure, and adapt to internal or external<br />
changes.<br />
Refers to <strong>the</strong> willingness to make decisions, demonstrating initiative, and<br />
maintaining one’s position until convinced o<strong>the</strong>rwise by facts.<br />
Refers to <strong>the</strong> overall need for interaction and coordination among crew<br />
members<br />
Domain <strong>of</strong> application<br />
The CDA technique is generic and can be applied to any task that involves teamwork or<br />
collaboration.<br />
39
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Procedure and advice<br />
Step 1: Define task(s) under analysis<br />
The first step in a CDA is to define <strong>the</strong> task or scenario that will be analysed. This is<br />
dependent upon <strong>the</strong> focus <strong>of</strong> <strong>the</strong> analysis. It is recommended that if team coordination in<br />
a particular type <strong>of</strong> system (e.g. command and control) is under investigation, <strong>the</strong>n a set<br />
<strong>of</strong> scenarios representative <strong>of</strong> team performance in <strong>the</strong> system should be used. If time<br />
and financial constraints do not allow this, <strong>the</strong>n a task that is as representative as possible<br />
<strong>of</strong> team performance in <strong>the</strong> system under analysis should be used.<br />
Step 2: Select appropriate teamwork taxonomy<br />
Once <strong>the</strong> task(s) under analysis are defined, an appropriate teamwork taxonomy should<br />
be selected. Again, this may depend upon <strong>the</strong> purpose <strong>of</strong> <strong>the</strong> analysis. However, it is<br />
recommended that <strong>the</strong> taxonomy used covers all aspects <strong>of</strong> teamwork in <strong>the</strong> task under<br />
analysis.<br />
Step 3: Data collection phase<br />
Typically, data regarding <strong>the</strong> task(s) under analysis is collected using observation and/or<br />
interviews. Specific data regarding <strong>the</strong> task under analysis should be collected during<br />
this process, including information regarding each task step, each team member’s roles,<br />
and communications made. It is recommended that video and audio recording equipment<br />
are used to record any observations or interviews conducted during this process.<br />
Step 4: Conduct a HTA for <strong>the</strong> task under analysis<br />
Once sufficient data regarding <strong>the</strong> task under analysis has been collected, a HTA should<br />
be conducted. The HTA should represent a breakdown <strong>of</strong> <strong>the</strong> task under analysis in<br />
terms <strong>of</strong> goals, sub-goals, operations and plans.<br />
Step 5: Construct CDA rating sheet<br />
Once a HTA for <strong>the</strong> task under analysis is completed, a CDA rating sheet should be<br />
created. The rating sheet should include a column containing each bottom level task step<br />
as identified by <strong>the</strong> HTA. The teamwork behaviours from <strong>the</strong> taxonomy should run<br />
across <strong>the</strong> top if <strong>the</strong> table. An extract <strong>of</strong> a CDA rating sheet is presented in Table 17.<br />
Step 6: Taskwork/<strong>Teamwork</strong> classification<br />
Once <strong>the</strong> CDA rating sheet is complete, <strong>the</strong> analyst(s) should <strong>the</strong>n determine which <strong>of</strong> <strong>the</strong><br />
bottom level task steps from <strong>the</strong> HTA involve taskwork and which involve teamwork.<br />
Only those task steps defined as teamwork task steps should be rated according to <strong>the</strong><br />
teamwork taxonomy. Those task steps that are conducted individually involving no<br />
collaboration are classified as taskwork, whilst those task steps that are conducted<br />
collaboratively, involving more than one agent are classified as teamwork.<br />
Step 7: SME rating phase<br />
40
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Appropriate SME’s should <strong>the</strong>n rate <strong>the</strong> extent to which each teamwork behaviour is<br />
required during <strong>the</strong> completion <strong>of</strong> each teamwork task step. This involves presenting <strong>the</strong><br />
task step in question and discussing <strong>the</strong> role <strong>of</strong> each taxonomy behaviour in <strong>the</strong><br />
completion <strong>of</strong> <strong>the</strong> task step. An appropriate rating scale should be used e.g. low (1),<br />
medium (2) and high (3).<br />
Step 8: Calculate summary statistics<br />
Once all <strong>of</strong> <strong>the</strong> teamwork task steps have been rated according to <strong>the</strong> teamwork<br />
taxonomy, <strong>the</strong> final step is to calculate appropriate summary statistics. In its present<br />
usage, a total co-ordination value and mean co-ordination value for each teamwork task<br />
step are calculated. The mean co-ordination is simply an average <strong>of</strong> <strong>the</strong> ratings for <strong>the</strong><br />
teamwork behaviours for <strong>the</strong> task step in question. A mean overall co-ordination value<br />
for <strong>the</strong> entire scenario is also calculated.<br />
Example<br />
The following example is an extract <strong>of</strong> a CDA analysis <strong>of</strong> an energy distribution task<br />
(Salmon et al 2004b). The task involved <strong>the</strong> switching out <strong>of</strong> three circuits (SGT5 and<br />
SGT1A and 1B) at Barking 275Kv, 132Kv and 33Kv Substations. Circuit SGT5 was<br />
being switched out for <strong>the</strong> installation <strong>of</strong> a new transformer for <strong>the</strong> nearby channel tunnel<br />
rail link and SGT1A and 1B were being switched out for substation maintenance.<br />
Observational data from Barking substation and <strong>the</strong> NOC control room was used to<br />
conduct a HTA <strong>of</strong> <strong>the</strong> switching scenario. Each bottom level task in <strong>the</strong> HTA was <strong>the</strong>n<br />
defined by <strong>the</strong> analyst(s) as ei<strong>the</strong>r taskwork or teamwork. Each teamwork task was <strong>the</strong>n<br />
rated using <strong>the</strong> CDA taxonomy on a scale <strong>of</strong> 1 (low) to 3 (high). An extract <strong>of</strong> <strong>the</strong> HTA<br />
for <strong>the</strong> task is presented in Figure 3-11. An extract <strong>of</strong> <strong>the</strong> CDA is presented in Table 17.<br />
The overall CDA results are presented in Table 18.<br />
NGC Switching operations HTA<br />
0. Co-ordinate and carry out switching operations on circuits SGT5. SGT1A and 1B at<br />
Bark s/s (Plan 0. Do 1 <strong>the</strong>n 2 <strong>the</strong>n 3, EXIT)<br />
1. Prepare for switching operations (Plan 1. Do 1.1, <strong>the</strong>n 1.2, <strong>the</strong>n 1.3, <strong>the</strong>n 1.4, <strong>the</strong>n 1.5,<br />
<strong>the</strong>n 1.6, <strong>the</strong>n 1.7, <strong>the</strong>n 1.8, <strong>the</strong>n 1.9,<strong>the</strong>n 1.10 EXIT)<br />
1.1. Agree SSC (Plan 1.1. Do 1.1.1, <strong>the</strong>n 1.1.2, <strong>the</strong>n 1.1.3, <strong>the</strong>n 1.1.4, <strong>the</strong>n 1.1.5, EXIT)<br />
1.1.1. (WOK) Use phone to Contact NOC<br />
1.1.2. (WOK + NOC) Exchange identities<br />
1.1.3. (WOK + NOC) Agree SSC documentation<br />
1.1.4. (WOK+NOC) Agree SSC and time (Plan 1.1.4. Do 1.1.4.1, <strong>the</strong>n 1.1.4.2,<br />
EXIT)<br />
1.1.4.1. (NOC) Agree SSC with WOK<br />
41
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
1.1.4.2. (NOC) Agree time with WOK<br />
1.1.5. (NOC) Record and enter details (Plan 1.1.5. Do 1.1.5.1, <strong>the</strong>n 1.1.5.2, EXIT)<br />
1.1.5.1. Record details on log sheet<br />
1.1.5.2. Enter details into worksafe<br />
1.2. (NOC) Request remote isolation (Plan 1.2. Do 1.2.1, <strong>the</strong>n 1.2.2, <strong>the</strong>n 1.2.3, <strong>the</strong>n<br />
1.2.4, EXIT)<br />
1.2.1. (NOC) Ask WOK for isolators to be opened remotely<br />
1.2.2. (WOK) Perform remote isolation<br />
1.2.3. (NOC) Check Barking s/s screen<br />
1.2.4. (WOK + NOC) End communications<br />
1.3. Ga<strong>the</strong>r information on outage at transformer 5 at Bark s/s<br />
(Plan 1.3. Do 1.3.1, <strong>the</strong>n 1.3.2, <strong>the</strong>n 1.3.3, <strong>the</strong>n 1.3.4, EXIT)<br />
1.3.1. (NOC) Use phone to contact SAP at Bark<br />
1.3.2. (NOC + SAP) Exchange identities<br />
Figure 3-11 - Extract <strong>of</strong> HTA for NGT Switching scenario<br />
42
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 17 - CDA extract (Source: Salmon et al 2004)<br />
Task Step Agent Step No. Task<br />
Work<br />
Team<br />
Work<br />
Comm SA DM MA Lead Ad Ass TOT<br />
CO-ORD<br />
Mode<br />
TOT<br />
CO-ORD<br />
Mean<br />
1.1.1 WOK control room Use phone to contact NOC 1<br />
operator<br />
1.1.2 WOK control room Exchange identities 1 3 3 1 1 1 1 1 1.00 1.57<br />
operator<br />
NOC control room operator<br />
1.1.3 WOK control room Agree SSC documentation 1 3 3 3 1 1 1 1 1.00 1.86<br />
operator<br />
NOC control room operator<br />
1.1.4.1 NOC control room Agree SSC with WOK 1 3 3 3 1 1 1 1 1.00 1.86<br />
operator<br />
1.1.4.2 NOC control room Agree time with WOK 1 3 3 3 1 1 1 1 1.00 1.86<br />
operator<br />
1.1.5.1 NOC control room Record details onto log sheet 1<br />
operator<br />
1.1.5.2 NOC control room Enter details into WorkSafe 1<br />
operator<br />
1.2.1 NOC control room Ask for isolators to be opened remotely 1 3 3 1 2 2 1 1 1.00 1.86<br />
operator<br />
1.2.2 WOK control room Perform remote isolation 1<br />
operator<br />
1.2.3 NOC control room Check barking s/s screen 1<br />
operator<br />
1.2.4 WOK control room End communications 1 3 1 1 1 1 1 1 1.00 1.29<br />
operator<br />
NOC control room operator<br />
1.3.1 NOC control room Use phone to contact SAP at Barking 1<br />
operator<br />
1.3.2 NOC control room Exchange identities 1 3 3 1 1 1 1 1 1.00 1.57<br />
operator<br />
SAP at Barking<br />
43
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 18 - CDA results (Source: Salmon et al 2004)<br />
Category<br />
Result<br />
Total task steps 314<br />
Total taskwork 114 (36%)<br />
Total teamwork 200 (64%)<br />
Mean Total Co-ordination 1.57<br />
Modal Total Co-ordination 1.00<br />
Minimum Co-ordination 1.00<br />
Maximum Co-ordination 2.14<br />
Advantages<br />
Disadvantages<br />
• Offers a rating <strong>of</strong> co-ordination between team members for each teamwork<br />
based task step.<br />
• The output <strong>of</strong> a CDA is very useful, <strong>of</strong>fering an insight into <strong>the</strong> use <strong>of</strong><br />
teamwork behaviours and also a rating <strong>of</strong> team coordination and its<br />
components.<br />
• CDA is particularly useful for <strong>the</strong> analysis <strong>of</strong> C4i activity.<br />
• The teamwork taxonomy presented by Burke (2004) covers all aspects <strong>of</strong><br />
team performance and coordination. The taxonomy is also generic, allowing<br />
<strong>the</strong> technique to be used in any domain without modification.<br />
• CDA provides a breakdown <strong>of</strong> team performance in terms <strong>of</strong> task steps and<br />
<strong>the</strong> level <strong>of</strong> co-ordination required.<br />
• The technique is generic and can be applied to teamwork scenarios in any<br />
domain.<br />
• The CDA rating procedure is time consuming and laborious. The initial<br />
data collection phase and <strong>the</strong> creation <strong>of</strong> a HTA for <strong>the</strong> task under analysis<br />
also add fur<strong>the</strong>r time to <strong>the</strong> analysis.<br />
44
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Related methods<br />
• For <strong>the</strong> technique to be used properly, <strong>the</strong> appropriate SME’s are required.<br />
It may be difficult to gain access to SME’s for <strong>the</strong> required period <strong>of</strong> time.<br />
• Intra analyst and inter analyst reliability is questionable. Different SME’s<br />
may <strong>of</strong>fer different teamwork ratings for <strong>the</strong> same task (intra analyst<br />
reliability), whilst SME’s may provide different ratings on different<br />
occasions.<br />
In conducting a CDA analysis, a number <strong>of</strong> o<strong>the</strong>r HF techniques are used. Data regarding<br />
<strong>the</strong> task under analysis is typically collected using observations and interviews. A HTA<br />
for <strong>the</strong> task under analysis is normally conducted, <strong>the</strong> output <strong>of</strong> which feeds into <strong>the</strong> CDA<br />
rating sheet. A likert style rating scale is also normally used during <strong>the</strong> team behaviour<br />
rating procedure. Burke (2004) also suggests that a CDA should be conducted as part <strong>of</strong><br />
an overall team task analysis procedure. The CDA technique has also recently been<br />
integrated with a number <strong>of</strong> o<strong>the</strong>r techniques (HTA, observation, comms usage diagram,<br />
social network analysis, operator sequence diagrams and propositional networks) to form<br />
<strong>the</strong> event analysis <strong>of</strong> systemic teamwork (EAST) methodology, which has been used to<br />
analyse C4i activity in a number <strong>of</strong> domains.<br />
Approximate training and application times<br />
The training time for <strong>the</strong> CDA technique is minimal, requiring only that <strong>the</strong> SME’s used<br />
understand each <strong>of</strong> <strong>the</strong> behaviours specified in <strong>the</strong> teamwork taxonomy and also <strong>the</strong><br />
rating procedure. The application time is medium to high, depending upon <strong>the</strong> size and<br />
complexity <strong>of</strong> <strong>the</strong> scenario under analysis. In <strong>the</strong> CDA provided in <strong>the</strong> analysis, <strong>the</strong><br />
ratings procedure alone took approximately four hours. This represents a medium<br />
application time.<br />
Reliability and validity<br />
There are no data regarding <strong>the</strong> reliability and validity <strong>of</strong> <strong>the</strong> technique available in <strong>the</strong><br />
literature. Certainly both <strong>the</strong> intra analyst and inter analyst reliability <strong>of</strong> <strong>the</strong> technique<br />
may be questionable, and this may be dependent upon <strong>the</strong> type <strong>of</strong> rating scale used e.g. it<br />
is estimated that <strong>the</strong> reliability may be low when using a scale <strong>of</strong> 1-10, whilst it may be<br />
improved using a scale <strong>of</strong> one to three (low to high).<br />
Tools needed<br />
During <strong>the</strong> data collection phase, video (e.g. camcorder) and audio (e.g. recordable minidisc<br />
player) recording equipment are required in order to make a recording <strong>of</strong> <strong>the</strong> task or<br />
scenario under analysis. Once <strong>the</strong> data collection phase is complete, <strong>the</strong> CDA technique<br />
can be conducted using pen and paper.<br />
Flowchart<br />
45
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
START<br />
Define <strong>the</strong> task(s)<br />
under analysis<br />
Data collection phase<br />
Conduct a HTA for <strong>the</strong><br />
task under analysis<br />
Create CDA rating<br />
sheet<br />
Take <strong>the</strong> first/next step<br />
on <strong>the</strong> rating sheet<br />
Rate following behaviours on<br />
scale <strong>of</strong> 1 (low) to 3 (high):<br />
• Communication<br />
• Situation<br />
awareness<br />
• Decision making<br />
• Mission analysis<br />
• Leadership<br />
• Adaptability<br />
• Assertiveness<br />
Calculate <strong>the</strong> total<br />
coordination involved<br />
in <strong>the</strong> task step<br />
Are<br />
<strong>the</strong>re any<br />
more<br />
STOP<br />
46
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
3.5 CUD – Comms Usage Diagram<br />
Background and applications<br />
Comms Usage Diagram (CUD) (Watts & Monk 2000) is used to describe collaborative<br />
activity between teams <strong>of</strong> agents dispersed across different geographical locations. A<br />
CUD output describes <strong>the</strong> order <strong>of</strong> communications during <strong>the</strong> scenario under analysis,<br />
how and why communications between agents occur, which technology is involved in <strong>the</strong><br />
communication, and <strong>the</strong> advantages and disadvantages <strong>of</strong> <strong>the</strong> technology used. The CUD<br />
technique was originally developed and applied in <strong>the</strong> telecommunications domain,<br />
whereby <strong>the</strong> technique was used to analyse ‘telemedical consultation’ involving a<br />
medical practitioner <strong>of</strong>fering advice regarding a medical ailment from a different location<br />
to <strong>the</strong> advice seeker (Watts & Monk 2000). The technique has more recently been used<br />
by <strong>the</strong> authors in <strong>the</strong> analysis <strong>of</strong> C4i activity in a number <strong>of</strong> domains, including energy<br />
distribution, naval warfare, fire services, air traffic control, military, rail and aviation<br />
domains (see example section). A CUD analysis is typically based upon observational<br />
data <strong>of</strong> <strong>the</strong> task or scenario under analysis, although talk through analysis and interviews<br />
are also used (Watts and Monk 2000).<br />
Domain <strong>of</strong> application<br />
Although <strong>the</strong> technique was originally developed for use in <strong>the</strong> medical domain, it is<br />
generic and can be applied in any domain that involves distributed collaboration.<br />
Procedure and advice<br />
Step 1: Define <strong>the</strong> task or scenario under analysis<br />
The first step in a CUD analysis is to clearly define <strong>the</strong> task or scenario under analysis. It<br />
may be useful to conduct a HTA <strong>of</strong> <strong>the</strong> task under analysis for this purpose. A clear<br />
definition <strong>of</strong> <strong>the</strong> task under analysis allows <strong>the</strong> analyst(s) to prepare for <strong>the</strong> data<br />
collection phase.<br />
Step 2: Data collection<br />
Next, <strong>the</strong> analyst(s) should collect specific data regarding <strong>the</strong> task or scenario under<br />
analysis. Watts & Monk (2000) recommend that interviews, observations and task talkthrough<br />
should be used to collect <strong>the</strong> data. Specific data regarding <strong>the</strong> personnel<br />
involved, activity, task steps, communication between personnel, technology used and<br />
geographical location should be collected.<br />
Step 3: Complete Initial Comms report<br />
Following <strong>the</strong> data collection phase, <strong>the</strong> raw data obtained should be put into a report<br />
form. According to Watts & Monk (2000) <strong>the</strong> report should include <strong>the</strong> location <strong>of</strong> <strong>the</strong><br />
technology used, <strong>the</strong> purpose <strong>of</strong> <strong>the</strong> technology, <strong>the</strong> advantages and disadvantages <strong>of</strong><br />
using such technology, and graphical account <strong>of</strong> a typical consultation session. The<br />
report should <strong>the</strong>n be made available to all personnel involved for evaluation and<br />
reiteration purposes.<br />
47
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Step 4: Construct CUD output table<br />
The graphical account developed in step 3 forms <strong>the</strong> basis for <strong>the</strong> CUD output. First <strong>of</strong><br />
all, <strong>the</strong> analyst should construct an appropriate CUD template, containing a column for<br />
each agent involved in <strong>the</strong> scenario, and columns for <strong>the</strong> technology used, <strong>the</strong> advantages<br />
and disadvantages associated with <strong>the</strong> technology used and <strong>the</strong> recommended technology.<br />
An example CUD template is presented in Figure 3-12. The CUD output contains a<br />
description <strong>of</strong> <strong>the</strong> task activity at each geographical location and <strong>the</strong> collaboration<br />
between personnel at each location. Directional arrows (i.e. from/to) should <strong>the</strong>n be used<br />
to represent <strong>the</strong> communications between personnel at different locations. The comms<br />
column specifies <strong>the</strong> technology used in <strong>the</strong> communication and <strong>the</strong> effects column lists<br />
any advantages and disadvantages associated with <strong>the</strong> technology in question. This is<br />
based upon <strong>the</strong> initial observation <strong>of</strong> <strong>the</strong> scenario and also analyst subjective judgement.<br />
Finally, <strong>the</strong> recommended comms column is used to provide a recommendation <strong>of</strong> <strong>the</strong><br />
most suitable technology available for <strong>the</strong> comms in question.<br />
NGT Feckenham Switching Operations 12 th July 2004<br />
Com ms Usage Diagram<br />
TIME<br />
NOC control room<br />
operator<br />
SAP/AP at<br />
Tottenham<br />
Substation<br />
SAP/AP at<br />
W altham C ross<br />
Substation<br />
SAP/AP at<br />
Brimsdown<br />
substation<br />
Overhead Line<br />
Party contact<br />
W OK control<br />
room operator<br />
Comms<br />
Effects <strong>of</strong> comms<br />
medium used<br />
Recommended<br />
comms<br />
14:48<br />
15:00<br />
15:03<br />
15:32<br />
Step 5: Calculate comms technology figures<br />
Figure 3-12 - CUD template<br />
In order to summarise <strong>the</strong> CUD analysis, <strong>the</strong> frequency <strong>of</strong> communications made by each<br />
agent and <strong>the</strong> comms technology used is calculated. An example CUD summary table is<br />
presented in Table 19.<br />
Advantages<br />
• The CUD output is particularly useful, <strong>of</strong>fering a description <strong>of</strong> <strong>the</strong> task<br />
under analysis, and also a description <strong>of</strong> collaborative activity involved,<br />
including <strong>the</strong> order <strong>of</strong> activity, <strong>the</strong> communications between agents, <strong>the</strong><br />
personnel involved, <strong>the</strong> technology used and its associated advantages and<br />
disadvantages, and recommendations regarding <strong>the</strong> technology used.<br />
48
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Disadvantages<br />
Example<br />
• The output <strong>of</strong> a CUD analysis is particularly useful for highlighting<br />
communication flaws in a particular network <strong>of</strong> agents.<br />
• The CUD technique is particularly suited to <strong>the</strong> analysis <strong>of</strong> teamwork and<br />
C4i activity.<br />
• The CUD technique is also flexible, and could potentially be modified to<br />
make it more comprehensive. Factors such as time, error and workload<br />
could be incorporated into a CUD analysis, ensuring a much more<br />
exhaustive analysis.<br />
• Although <strong>the</strong> CUD technique was developed and originally used in <strong>the</strong><br />
medical domain, it is a generic technique and could potentially be applied in<br />
any domain involving distributed collaboration.<br />
• The technique is easy to use and quick to learn.<br />
• For large, complex tasks involving many agents, conducting a CUD<br />
analysis may become a lengthy and laborious process.<br />
• In its present usage, <strong>the</strong> CUD analysis only defines <strong>the</strong> communications<br />
technology used at <strong>the</strong> source <strong>of</strong> <strong>the</strong> communication in question (i.e. <strong>the</strong><br />
comms resource at <strong>the</strong> o<strong>the</strong>r end <strong>of</strong> <strong>the</strong> communication in question is not<br />
defined).<br />
• The initial data collection phase <strong>of</strong> <strong>the</strong> CUD technique is time consuming<br />
and labour intensive.<br />
• No validity or reliability data are available for <strong>the</strong> technique.<br />
• Application <strong>of</strong> <strong>the</strong> CUD technique appears to be limited.<br />
• No guidelines are <strong>of</strong>fered to <strong>the</strong> analyst in analysing <strong>the</strong> technology used in<br />
<strong>the</strong> communications under analysis. Typically this is based upon <strong>the</strong><br />
analyst’s subjective judgement. This may affect <strong>the</strong> reliability <strong>of</strong> <strong>the</strong><br />
technique.<br />
The following example is a CUD analysis <strong>of</strong> an energy distribution task. The task<br />
involved <strong>the</strong> return from isolation <strong>of</strong> <strong>the</strong> Brimsdown -Tottenham-Waltham Cross Circuit<br />
(Salmon et al 2004c). The data collection phase involved two observers. The first<br />
observer was situated at <strong>the</strong> National Grid Transco (NGT) National Operations Centre<br />
(NOC) in Warwick and observed <strong>the</strong> activity <strong>of</strong> <strong>the</strong> NOC control room operator. The<br />
second observer was situated at <strong>the</strong> Tottenham substation and observed <strong>the</strong> activity <strong>of</strong> <strong>the</strong><br />
senior authorised person (SAP) and authorised person (AP) who completed work required<br />
to return <strong>the</strong> circuit from isolation. The CUD analysis used observational data and a HTA<br />
49
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
<strong>of</strong> <strong>the</strong> task as its primary inputs. An extract <strong>of</strong> <strong>the</strong> CUD analysis for <strong>the</strong> energy<br />
distribution task is presented in Figure 3-13. In order to summarise <strong>the</strong> CUD analysis,<br />
comms frequency figures were calculated. The CUD summary results are presented in<br />
Table 19.<br />
NGT Feckenham Switching Operations 12 th July 2004<br />
TIME<br />
NOC control room<br />
operator<br />
SAP/AP at<br />
Tottenham<br />
Substation<br />
SAP/AP at<br />
W altham Cross<br />
S ubstation<br />
SAP/AP at<br />
Brimsdown<br />
substation<br />
Overhead Line<br />
Party contact<br />
WOK control<br />
room operator<br />
Comms<br />
Comms Usage Diagram<br />
Effects <strong>of</strong> comms medium<br />
used<br />
R ecomm ended<br />
comms<br />
08:50<br />
C ontact N OC<br />
control room<br />
operator to ask<br />
w hen deearthing<br />
ops can<br />
begin<br />
+ Sound is clear and stable<br />
+ Comms is two way<br />
+ Can look at site diagram /<br />
desktop screen at <strong>the</strong> sam e<br />
time<br />
+ Confidentiality maintained<br />
- D ata is not recorded.<br />
Information may be lost<br />
09:04<br />
Contact SAP to<br />
issue first<br />
instructions<br />
+ Sound is clear and stable<br />
+ Comms is two way<br />
+ Can look at site diagram /<br />
desktop screen at <strong>the</strong> sam e<br />
time<br />
+ Confidentiality maintained<br />
- D ata is not recorded.<br />
Information may be lost<br />
09:05<br />
Contact SAP to<br />
discuss<br />
forthcoming<br />
operations<br />
+ Sound is clear and stable<br />
+ Comms is two way<br />
+ Can look at site diagram /<br />
desktop screen at <strong>the</strong> sam e<br />
time<br />
+ Confidentiality maintained<br />
- D ata is not recorded.<br />
Information may be lost<br />
09:45<br />
Contact NOC to<br />
report las t<br />
instructions<br />
c omplete and to<br />
ask for next<br />
instructions<br />
+ Sound is clear and stable<br />
+ Comms is two way<br />
+ Can look at site diagram /<br />
desktop screen at <strong>the</strong> sam e<br />
time<br />
+ Confidentiality maintained<br />
- D ata is not recorded.<br />
Information may be lost<br />
10:00<br />
C ontact OLP to<br />
discuss<br />
forthcoming<br />
operations<br />
+ Sound is clear and stable<br />
+ Comms is two way<br />
+ Can look at site diagram /<br />
desktop screen at <strong>the</strong> sam e<br />
time<br />
+ Confidentiality maintained<br />
- D ata is not recorded.<br />
Information may be lost<br />
Figure 3-13 - Comms Usage Diagram for an energy distribution task (Salmon et al<br />
2004)<br />
Table 19 - CUD summary table<br />
Agent<br />
Telephon<br />
e comms<br />
Mobile phone<br />
comms<br />
PC comms<br />
Total<br />
NOC operator 11 11<br />
SAP/AP<br />
Tottenham<br />
SAP/AP<br />
Waltham Cross<br />
SAP/AP<br />
Brimsdown<br />
at<br />
at<br />
at<br />
8 1 9<br />
5 5<br />
7 7<br />
Overhead<br />
party contact<br />
line<br />
1 1<br />
WOK operator 1 1<br />
Total 32 2 34<br />
50
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Related methods<br />
During <strong>the</strong> data collection phase, a number <strong>of</strong> different techniques may be used, such as<br />
observational analysis, interviews and talk-through type analysis. It is also useful to<br />
conduct a HTA <strong>of</strong> <strong>the</strong> task under analysis prior to constructing <strong>the</strong> CUD. The CUD<br />
technique has also recently been integrated with a number <strong>of</strong> o<strong>the</strong>r techniques (HTA,<br />
observation, co-ordination demands analysis, social network analysis, operator sequence<br />
diagrams and propositional networks) to form <strong>the</strong> event analysis <strong>of</strong> systemic teamwork<br />
(EAST) methodology, which has been used to analyse C4i activity.<br />
Flowchart<br />
START<br />
Define <strong>the</strong> task or scenario<br />
under analysis<br />
Use observation and<br />
interviews to collect<br />
specific task data<br />
Transcribe <strong>the</strong> raw data<br />
into report form<br />
Construct comms usage<br />
diagram<br />
Calculate comms and<br />
technology frequency<br />
figures<br />
STOP<br />
Approximate training and application times<br />
The training time associated with CUD is very low, assuming that <strong>the</strong> practitioner was<br />
already pr<strong>of</strong>icient in data collection techniques such as interviews and observational<br />
analysis. In a recent training programme for <strong>the</strong> EAST methodology, <strong>the</strong> CUD training<br />
session lasted under 1 hour. The application time associated with <strong>the</strong> CUD technique is<br />
also typically low to medium, depending upon task size and complexity. In <strong>the</strong> analyse <strong>of</strong><br />
51
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
energy distribution scenarios (Salmon et al 2004a, b, c) <strong>the</strong> application time for CUD<br />
ranged between 1 and 4 hours, depending upon <strong>the</strong> size and complexity <strong>of</strong> <strong>the</strong> scenario<br />
under analysis.<br />
Reliability and validity<br />
No data regarding <strong>the</strong> reliability and validity <strong>of</strong> <strong>the</strong> technique are available.<br />
Tools needed<br />
A CUD analysis can be conducted using pen and paper only. However, it is<br />
recommended that video and audio recording equipment are used during <strong>the</strong> data<br />
collection phase. The CUD can be constructed using Micros<strong>of</strong>t Visio.<br />
3.6 Social Network <strong>Analysis</strong><br />
Background and applications<br />
Social Network <strong>Analysis</strong> (SNA) is used to analyse and represent <strong>the</strong> relationships<br />
between teams or agents within a social network. A social network is defined as a set or<br />
team <strong>of</strong> agents that possess relationships with one ano<strong>the</strong>r (Driskell & Mullen 2004).<br />
SNA is based upon <strong>the</strong> notion that <strong>the</strong> relationship between agents within a social<br />
network has a significant effect upon <strong>the</strong> actions performed and also <strong>the</strong> performance<br />
achieved by <strong>the</strong> network. SNA uses both graphical and ma<strong>the</strong>matical procedures to<br />
represent social networks. Typically, centrality measures are calculated for each agent<br />
(e.g. degree, betweenness and closeness) and <strong>the</strong> overall network density is calculated.<br />
This allows <strong>the</strong> identification <strong>of</strong> <strong>the</strong> key agents within <strong>the</strong> network and also <strong>the</strong><br />
classification <strong>of</strong> <strong>the</strong> network structure. The technique has previously been used for <strong>the</strong><br />
analysis <strong>of</strong> networks in a number <strong>of</strong> areas, including military C4ISR architectures<br />
(Dekker 2002), C4i activity in <strong>the</strong> fire service (Baber et al 2004), energy distribution<br />
(Salmon et al 2004), air traffic control, rail (Walker et al 2004), naval warfare (Stewart et<br />
al 2004) and military aviation domains, counselling and computer-mediated<br />
communications (Dekker 2002).<br />
Domain <strong>of</strong> application<br />
Generic.<br />
Procedure and advice<br />
Step 1: Define network or group<br />
The first step in a SNA involves defining <strong>the</strong> network or group <strong>of</strong> networks that are to be<br />
analysed. The type <strong>of</strong> network to be analysed should be specified first. For example, for<br />
work package 1.1, <strong>the</strong> authors specified C4i networks as <strong>the</strong> focus <strong>of</strong> <strong>the</strong>ir analysis. The<br />
purpose <strong>of</strong> this was <strong>the</strong> evaluation <strong>of</strong> C4i activity with a view to developing a generic<br />
model <strong>of</strong> C4i. Once <strong>the</strong> overall network type is specified, fur<strong>the</strong>r social networks within<br />
<strong>the</strong> network type specified should be defined. For <strong>the</strong> analysis <strong>of</strong> C4i networks, <strong>the</strong><br />
authors identified a number <strong>of</strong> different C4i networks that could be analysed. These were<br />
52
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
C4i networks within <strong>the</strong> emergency services, rail, military, aviation, nuclear reprocessing,<br />
air traffic control, civil energy distribution, and naval domains.<br />
Step 2: Define scenarios<br />
Typically, networks are analysed over a number <strong>of</strong> different scenarios. Once <strong>the</strong> type <strong>of</strong><br />
network under analysis has been defined, <strong>the</strong> scenario(s) within which <strong>the</strong>y will be<br />
analysed should be defined. For a thorough analysis, a number <strong>of</strong> different scenarios<br />
should be analysed. The scenarios used should be representative <strong>of</strong> all C4i activity within<br />
<strong>the</strong> network under analysis. In <strong>the</strong> analysis <strong>of</strong> naval warfare C4i (Stewart et al 2004), <strong>the</strong><br />
following scenarios were analysed:<br />
• Air threat scenario<br />
• Surface threat scenario<br />
• Subsurface threat scenario<br />
Step 3: Data collection<br />
Once <strong>the</strong> network and scenario(s) under analysis are defined clearly, <strong>the</strong> data collection<br />
phase can begin. The data collection phase involves <strong>the</strong> collection <strong>of</strong> specific data on <strong>the</strong><br />
relationship (e.g. communications and activity) between <strong>the</strong> agents during <strong>the</strong> scenario.<br />
Typical HF data collection techniques should be used in this process, such as<br />
observational analysis, interviews and questionnaires. Typically agent activity and <strong>the</strong><br />
frequency, direction and content <strong>of</strong> any communications between agents in <strong>the</strong> network<br />
are recorded.<br />
Step 4: Construct agent association matrix<br />
Once sufficient data regarding <strong>the</strong> scenario under analysis is collected, an agent<br />
association matrix should be constructed. The matrix represents <strong>the</strong> frequency <strong>of</strong><br />
associations between each agent in <strong>the</strong> network.<br />
Step 5: Construct social network diagram<br />
Once <strong>the</strong> matrix <strong>of</strong> association is completed, <strong>the</strong> social network diagram should be<br />
created. The social network depicts each agent in <strong>the</strong> network and <strong>the</strong> communications<br />
between <strong>the</strong>m. The communications are represented by directional arrows, and <strong>the</strong><br />
frequency <strong>of</strong> communications is also presented. An example social network diagram is<br />
presented in Figure 3-14.<br />
Step 6: Calculate agent centrality<br />
Agent centrality is calculated in order to determine <strong>the</strong> central or key agent(s) within <strong>the</strong><br />
network. There are a number <strong>of</strong> different centrality calculations that can be made. For<br />
example, agent centrality can be calculated using Bavelas-Leavitt’s index. The mean<br />
centrality + standard deviation can <strong>the</strong>n be used to define key agents within <strong>the</strong> network.<br />
Those agents who posses a centrality figure that exceeds <strong>the</strong> mean + standard deviation<br />
figure are defined as <strong>the</strong> key agents within <strong>the</strong> network.<br />
53
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Step 7: Calculate sociometric status<br />
The sociometric status <strong>of</strong> each agent refers to <strong>the</strong> number <strong>of</strong> communications received<br />
and emitted, relative to <strong>the</strong> number <strong>of</strong> nodes in <strong>the</strong> network. The mean sociometric status<br />
+ standard deviation can also be used to define key agents within <strong>the</strong> network. Those<br />
agents who posses a sociometric status figure that exceeds <strong>the</strong> mean + standard deviation<br />
figure can be defined as <strong>the</strong> key agents within <strong>the</strong> network.<br />
Step 8: Calculate network density<br />
Network density is equal to <strong>the</strong> total number <strong>of</strong> links between <strong>the</strong> agents in <strong>the</strong> network<br />
divided by <strong>the</strong> total number <strong>of</strong> possible links. Low network density figures are indicative<br />
<strong>of</strong> a well distributed network <strong>of</strong> agents. High density figures indicate a network that is not<br />
well distributed.<br />
Advantages<br />
• SNA can be used to determine <strong>the</strong> importance <strong>of</strong> different agents within a<br />
social network and also to classify <strong>the</strong> network type.<br />
• The SNA <strong>of</strong>fers a comprehensive analysis <strong>of</strong> <strong>the</strong> network in question. The<br />
key agents within <strong>the</strong> network are specified, as are <strong>the</strong> frequency and<br />
direction <strong>of</strong> communications within <strong>the</strong> network. Fur<strong>the</strong>r classifications<br />
include network type and network density. There are also additional<br />
analyses that can be calculated, such as betweenness, closeness and distance<br />
calculations.<br />
• Networks can be classified according to <strong>the</strong>ir structure. This is particularly<br />
useful when analysing networks across different domains.<br />
• SNA is particularly suited to <strong>the</strong> analysis <strong>of</strong> C4i scenarios.<br />
• The technique has been used extensively in <strong>the</strong> past for <strong>the</strong> analysis <strong>of</strong><br />
various social networks.<br />
• The technique is simple to learn and easy to use.<br />
• SNA is a very quick technique to apply. The provision <strong>of</strong> <strong>the</strong> Agna SNA<br />
s<strong>of</strong>tware package also reduces application time fur<strong>the</strong>r.<br />
• SNA is a generic technique that could potentially be applied in any domain.<br />
Disadvantages<br />
54
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Example<br />
• For large, complex networks, it may be difficult to conduct a SNA.<br />
Application time is a function <strong>of</strong> network size, and large networks may<br />
incur lengthy application times.<br />
• The data collection phase involved in a SNA is also resource intensive.<br />
• Some knowledge <strong>of</strong> ma<strong>the</strong>matical techniques is required.<br />
• It is difficult to collect comprehensive data for a SNA. For example, a<br />
dispersed network <strong>of</strong> ten agents would require at least 10 observers in order<br />
to accurately capture <strong>the</strong> communications made between all agents.<br />
The following example is taken from an EAST analysis <strong>of</strong> a civil energy distribution<br />
scenario (Salmon et al 2004). The scenario involved <strong>the</strong> return to service <strong>of</strong> <strong>the</strong><br />
Brimsdown-Tottenham-Waltham Cross Circuit and took place at Brimsdown, Tottenham,<br />
Waltham Cross substations and <strong>the</strong> National Grid Transco (NGT) Network Operations<br />
Centre (NOC) in Warwick. The agents involved in <strong>the</strong> scenario are presented in Table 20.<br />
Table 20 - Agents involved in <strong>the</strong> return to service scenario<br />
Role <strong>of</strong> agent A NOC control room operator<br />
Role <strong>of</strong> agent B SAP/AP at Tottenham substation<br />
Role <strong>of</strong> agent C SAP/AP at Waltham Cross<br />
substation<br />
Role <strong>of</strong> agent D SAP/AP at Brimsdown substation<br />
Role <strong>of</strong> agent E Overhead line party contact<br />
Role <strong>of</strong> agent F WOK control room operator<br />
From <strong>the</strong> list <strong>of</strong> agents identified for <strong>the</strong> scenario, a matrix <strong>of</strong> association can be<br />
constructed. This matrix shows whe<strong>the</strong>r or not an agent within <strong>the</strong> system can be<br />
associated with any o<strong>the</strong>r agent, specifically through frequency <strong>of</strong> communications. The<br />
association matrix for <strong>the</strong> switching scenario is presented in Table 21.<br />
55
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 21 - Agent association matrix<br />
A B C D E F<br />
A 0 2 2 2 1 4<br />
B 8 0 0 0 0 1<br />
C 4 0 0 0 0 0<br />
D 8 0 0 0 0 0<br />
E 1 0 0 0 0 0<br />
F 0 1 0 0 0 0<br />
Finally, a social network diagram is created. The social network diagram illustrates <strong>the</strong><br />
proposed association between agents involved in <strong>the</strong> scenario. The numbers associated<br />
with <strong>the</strong> links between <strong>the</strong> agents in <strong>the</strong> system indicate <strong>the</strong> strength <strong>of</strong> association. The<br />
strength <strong>of</strong> association is defined by <strong>the</strong> number <strong>of</strong> occasions on which agents exchanged<br />
information. The direction <strong>of</strong> association is represented by directional arrows. The social<br />
network diagram is presented in Figure 3-14.<br />
Figure 3-14 - Return to service social network diagram.<br />
56
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
There are a number <strong>of</strong> ways to analyse social networks. In this case, agent centrality,<br />
sociometric status and network density were calculated. Agent centrality was calculated<br />
using Bavelas-Leavitt’s index. Table 22 shows <strong>the</strong> Centrality for <strong>the</strong> agents in this<br />
incident. The mean centrality was calculated as 3.13. A notion <strong>of</strong> ‘key’ agents can be<br />
defined using <strong>the</strong> mean + 1 standard deviation (i.e., 3.13 + 0.74 = 3.87). Using this rule,<br />
<strong>the</strong> B-L centrality calculation indicates that <strong>the</strong> NOC operator and <strong>the</strong> SAP/AP at<br />
Tottenham substation are <strong>the</strong> key agents in <strong>the</strong> network.<br />
Table 22 - Agent centrality (B-L Centrality)<br />
Agent<br />
B-L Centrality<br />
NOC operator 4.72<br />
SAP/AP at Tottenham 3.25<br />
SAP/AP at Waltham Cross 2.73<br />
SAP/AP at Brimsdown 2.73<br />
Overhead line party 2.73<br />
WOK operator 2.6<br />
Table 23 shows <strong>the</strong> Sociometric Status for each agent involved in <strong>the</strong> scenario. From <strong>the</strong><br />
calculation, a mean status <strong>of</strong> 2.26 (±3.82) was found. The value <strong>of</strong> mean + one standard<br />
deviation, i.e. 2.26 + 3.82 = 6.08, is used to define ‘key’ agents in this network. Again,<br />
<strong>the</strong> sociometric status analysis indicates that <strong>the</strong> NOC operator is <strong>the</strong> key agent within <strong>the</strong><br />
network.<br />
Table 23 - Agent sociometric status<br />
Agent<br />
Sociometric status<br />
NOC operator 6.4<br />
SAP/AP at Tottenham 2.4<br />
SAP/AP at Waltham Cross 1.2<br />
SAP/AP at Brimsdown 2.0<br />
Overhead line party 0.4<br />
WOK operator 1.2<br />
An overall measure <strong>of</strong> network density was also be derived by dividing <strong>the</strong> links actually<br />
present in <strong>the</strong> scenario, by all <strong>of</strong> <strong>the</strong> available links. For <strong>the</strong> Tottenham scenario, <strong>the</strong><br />
overall network density is calculated as 0.2 (6 links present divided by 30 possible links).<br />
This figure is suggestive <strong>of</strong> a well distributed, (and <strong>the</strong>refore less dense) network <strong>of</strong><br />
agents.<br />
57
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Related methods<br />
In terms <strong>of</strong> rating relationships in networks <strong>the</strong> SNA technique appears to be unique.<br />
SNA also uses ma<strong>the</strong>matical techniques for <strong>the</strong> analysis <strong>of</strong> <strong>the</strong> networks involved. During<br />
<strong>the</strong> data collection phase, techniques such as observational study, interviews and<br />
questionnaires are typically used.<br />
Approximate training and application times<br />
The associated training time for <strong>the</strong> SNA methodology is low. Although some knowledge<br />
<strong>of</strong> ma<strong>the</strong>matical analysis is required, <strong>the</strong> basic SNA procedure is a simple one. It is<br />
estimated that <strong>the</strong> technique could be trained to HF practitioners within 1 hour. The<br />
application time for <strong>the</strong> SNA technique is also minimal. The provision <strong>of</strong> Agna SNA<br />
s<strong>of</strong>tware package reduces <strong>the</strong> application time even fur<strong>the</strong>r. In a SNA <strong>of</strong> an energy<br />
distribution switching scenario (involving a network <strong>of</strong> four agents) <strong>the</strong> application time<br />
was approximately 2 hours. However, <strong>the</strong> data collection phase involved adds to <strong>the</strong><br />
overall application time, and this is typically lengthy. Also, <strong>the</strong> application time is<br />
dependent upon <strong>the</strong> size and complexity <strong>of</strong> <strong>the</strong> network under analysis, larger more<br />
complex networks may incur greater application times.<br />
Reliability and validity<br />
No data regarding <strong>the</strong> reliability and validity <strong>of</strong> <strong>the</strong> SNA technique are available.<br />
Tools needed<br />
SNA can be conducted using pen and paper, once <strong>the</strong> data collection phase is complete.<br />
However, <strong>the</strong>re are various SNA s<strong>of</strong>tware packages that can be used that automate <strong>the</strong><br />
SNA procedure. For example, based upon an association matrix as input, <strong>the</strong> AGNA<br />
s<strong>of</strong>tware package constructs <strong>the</strong> social network and also performs various statistical<br />
analyses <strong>of</strong> <strong>the</strong> network in question. Fur<strong>the</strong>r s<strong>of</strong>tware packages are available, such as<br />
UCINET (Borgatti, Everett, and Freeman, 2002) and STRUCTURE (Burt, 1991). The<br />
tools required during <strong>the</strong> data collection phase for a SNA would be dependent upon <strong>the</strong><br />
type <strong>of</strong> data collection techniques used. Observational analysis, interviews and<br />
questionnaires would normally require visual and audio recording equipment (video<br />
cameras, minidisc recorder, PC).<br />
58
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Flowchart<br />
START<br />
Define network to be<br />
analysed<br />
Define scenarios required<br />
Collect data for <strong>the</strong><br />
scenario(s) under analysis<br />
Construct association<br />
matrix<br />
Calculate <strong>the</strong> following as<br />
appropriate:<br />
• Centrality<br />
• Sociometric status<br />
• Network density<br />
• Betweenness<br />
• Distance<br />
Classify network<br />
STOP<br />
59
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
3.7 Operation Sequence Diagrams<br />
Background and applications<br />
Operation Sequence Diagrams (OSD) are used to graphically describe <strong>the</strong> activity and<br />
interaction between teams <strong>of</strong> agents within a network. According to Kirwan and<br />
Ainsworth (1992), <strong>the</strong> original purpose <strong>of</strong> OSD analysis was to represent complex, multiperson<br />
tasks. The output <strong>of</strong> an OSD graphically depicts <strong>the</strong> task process, including <strong>the</strong><br />
tasks performed and <strong>the</strong> interaction between operators over time, using standardised<br />
symbols. There are various forms <strong>of</strong> OSD’s, ranging from a simple flow diagram<br />
representing task order, to more complex OSD which account for team interaction and<br />
communication. OSD’s have been applied in a number <strong>of</strong> domains, including <strong>the</strong> fire<br />
service (Baber et al 2004), naval warfare (Stewart et al 2004), aviation (Stewart et al<br />
2004), energy distribution (Salmon et al 2004), air traffic control (Walker et al 2004) and<br />
rail (Walker et al 2004) domains.<br />
Domain <strong>of</strong> application<br />
The technique was originally used in <strong>the</strong> nuclear power and chemical process industries.<br />
However, <strong>the</strong> technique is generic and can be applied in any domain.<br />
Procedure and advice<br />
Step 1: Define <strong>the</strong> task(s) under analysis<br />
The first step in an OSD analysis is to define <strong>the</strong> task(s) or scenario(s) under analysis.<br />
The task(s) or scenario(s) should be defined clearly, including <strong>the</strong> activity and agents<br />
involved.<br />
Step 2: Data collection<br />
In order to construct an OSD, <strong>the</strong> analyst(s) must obtain sufficient data regarding <strong>the</strong> task<br />
or scenario under analysis. It is recommended that <strong>the</strong> analyst(s) use various forms <strong>of</strong><br />
data collection in this phase. Observational study should be used to observe <strong>the</strong> task (or<br />
similar types <strong>of</strong> task) under analysis. Interviews with personnel involved in <strong>the</strong> task (or<br />
similar tasks) should also be conducted. The type and amount <strong>of</strong> data collected in step 2<br />
is dependent upon <strong>the</strong> analysis requirements. The more exhaustive <strong>the</strong> analysis is<br />
intended to be, <strong>the</strong> more data collection techniques should be used.<br />
Step 3: Describe <strong>the</strong> task or scenario using HTA<br />
Once <strong>the</strong> data collection phase is completed, a detailed task analysis should be conducted<br />
for <strong>the</strong> scenario under analysis. The type <strong>of</strong> task analysis is determined by <strong>the</strong> analyst(s),<br />
and in some cases, a task list will suffice. However, it is recommended that a HTA is<br />
conducted.<br />
Step 4: Construct <strong>the</strong> OSD diagram<br />
60
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Once <strong>the</strong> task has been described adequately, <strong>the</strong> construction <strong>of</strong> <strong>the</strong> OSD can begin. The<br />
OSD template should include <strong>the</strong> title <strong>of</strong> <strong>the</strong> task step, a timeline, and a row for each<br />
agent involved in <strong>the</strong> task. An OSD template is presented in Figure 3-15. In order to<br />
construct <strong>the</strong> OSD, it is recommended that <strong>the</strong> analyst walks through <strong>the</strong> HTA <strong>of</strong> <strong>the</strong> task<br />
under analysis, constructing <strong>the</strong> OSD in conjunction. An OSD symbol glossary is<br />
presented in Figure 3-16. The symbols involved in a particular task step should be linked<br />
by directional arrows, in order to represent <strong>the</strong> flow <strong>of</strong> activity during <strong>the</strong> task. Each<br />
symbol in <strong>the</strong> OSD should contain <strong>the</strong> corresponding task step number from <strong>the</strong> HTA <strong>of</strong><br />
<strong>the</strong> task under analysis. The artefacts used during <strong>the</strong> communications should also be<br />
annotated onto <strong>the</strong> OSD.<br />
REC control<br />
room operator<br />
WOK Control<br />
Room<br />
Operator<br />
SAP/AP at<br />
Barking<br />
Substation<br />
NOC Control<br />
Room<br />
Operator<br />
Step 5: Overlay additional analyses results<br />
Figure 3-15 - Example OSD template<br />
One <strong>of</strong> <strong>the</strong> endearing features <strong>of</strong> <strong>the</strong> OSD technique is that additional analysis results can<br />
easily be added to <strong>the</strong> OSD. According to <strong>the</strong> analysis requirements, additional task<br />
features can also be annotated onto <strong>the</strong> OSD. For example, Baber et al (2004) added total<br />
co-ordination values for teamwork task steps (from an initial CDA analysis) onto <strong>the</strong><br />
OSD for a fire service task.<br />
Step 6: Calculate operation loading figures<br />
From <strong>the</strong> OSD, operational loading figures are calculated for each agent involved in <strong>the</strong><br />
scenario under analysis. Operational loading figures are calculated for each OSD operator<br />
or symbol used e.g. operation, receive, delay, decision, transport, and combined<br />
operations. The operational loading figures refer to <strong>the</strong> frequency in which each agent<br />
was involved in <strong>the</strong> operator in question during <strong>the</strong> scenario.<br />
Advantages<br />
61
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Disadvantages<br />
• The OSD provides an exhaustive analysis <strong>of</strong> <strong>the</strong> task in question. The flow<br />
<strong>of</strong> <strong>the</strong> task is represented in terms <strong>of</strong> activity and information, <strong>the</strong> type <strong>of</strong><br />
activity and <strong>the</strong> agents involved is specified, a timeline <strong>of</strong> <strong>the</strong> activity, <strong>the</strong><br />
communications between agents involved in <strong>the</strong> task, <strong>the</strong> technology used<br />
and also a rating <strong>of</strong> total co-ordination for each teamwork activity is also<br />
provided. The techniques flexibility also permits <strong>the</strong> analyst(s) to add<br />
fur<strong>the</strong>r analysis outputs onto <strong>the</strong> OSD, adding to its exhaustiveness.<br />
• An OSD is particularly useful for analysing and representing distributed<br />
teamwork or collaborated activity.<br />
• OSD’s are useful for demonstrating <strong>the</strong> relationship between tasks,<br />
technology and team members.<br />
• High face validity (Kirwan & Ainsworth 1992).<br />
• OSD’s have been used extensively in <strong>the</strong> past and have been applied in a<br />
variety <strong>of</strong> domains involving team based activity, including <strong>the</strong> emergency<br />
services (Baber et al 2004), energy distribution (Salmon et al 2004), rail<br />
(Walker et al 2004), air traffic control (ATC) (Walker et al 2004) and naval<br />
operations (Stewart et al 2004).<br />
• A number <strong>of</strong> different analyses can be overlaid onto an OSD <strong>of</strong> a particular<br />
task. For example, Baber et al (2004) add <strong>the</strong> corresponding HTA task step<br />
numbers and co-ordination demands analysis results to OSD’s <strong>of</strong> C4i<br />
activity.<br />
• The OSD technique is very flexible and can be modified to suit <strong>the</strong> analysis<br />
needs.<br />
• The WESTT s<strong>of</strong>tware package can be used to automate a large portion <strong>of</strong><br />
<strong>the</strong> OSD procedure.<br />
• The application time for an OSD analysis is lengthy. Constructing an OSD<br />
for large, complex tasks can be extremely time consuming and <strong>the</strong> initial<br />
data collection adds fur<strong>the</strong>r time to <strong>the</strong> analysis<br />
• The construction <strong>of</strong> large, complex OSD’s is also quite difficult.<br />
• OSD’s can become cluttered and confusing (Kirwan & Ainsworth 1992).<br />
• The output <strong>of</strong> OSD’s can become large and unwieldy.<br />
• The present OSD symbols are limited for certain applications (e.g. C4i<br />
scenarios).<br />
62
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
• The reliability <strong>of</strong> <strong>the</strong> technique is questionable. Different analysts may<br />
interpret <strong>the</strong> OSD symbols differently.<br />
Related methods<br />
Various types <strong>of</strong> OSD exist, including temporal operational sequence diagrams,<br />
partitioned operational sequence diagrams and spatial operational sequence diagrams<br />
(Kirwan & Ainsworth 1992). During <strong>the</strong> OSD data collection phase, techniques such as<br />
observational study and interviews are typically used. Task analysis techniques such as<br />
HTA are also used during <strong>the</strong> construction <strong>of</strong> <strong>the</strong> OSD. Timeline analysis may also be<br />
used in order to construct an appropriate timeline for <strong>the</strong> task or scenario under analysis.<br />
Additional analyses results can also be annotated onto an OSD, such as CDA and comms<br />
usage diagram. The OSD technique has also recently been integrated with a number <strong>of</strong><br />
o<strong>the</strong>r techniques (HTA, observation, co-ordination demands analysis, comms usage<br />
diagram, social network analysis and propositional networks) to form <strong>the</strong> event analysis<br />
<strong>of</strong> systemic teamwork (EAST) methodology (Baber et al 2004), which has been used to<br />
analyse C4i activity in a number <strong>of</strong> domains.<br />
Approximate training and application times<br />
The training time associated with <strong>the</strong> OSD methodology is low. However, <strong>the</strong> typical<br />
application time for <strong>the</strong> technique is high. The construction <strong>of</strong> an OSD can be a very time<br />
consuming process, involving many re-iterations. A typical OSD normally can take 12 +<br />
hours to construct.<br />
Reliability and validity<br />
According to Kirwan & Ainsworth (1992), OSD techniques possess a high degree <strong>of</strong> face<br />
validity. The intra analyst reliability <strong>of</strong> <strong>the</strong> technique may be suspect, as different<br />
analysts may interpret <strong>the</strong> OSD symbols differently.<br />
Tools needed<br />
When conducting an OSD analysis, pen and paper could be sufficient. However, to<br />
ensure that data collection is comprehensive, it is recommended that video or audio<br />
recording devices are used in conjunction with <strong>the</strong> pen and paper. For <strong>the</strong> construction <strong>of</strong><br />
<strong>the</strong> OSD, it is recommended that a suitable drawing package, such as Micros<strong>of</strong>t Visio<br />
is used. The WESTT s<strong>of</strong>tware package can also be used to automate a large portion <strong>of</strong> <strong>the</strong><br />
OSD procedure. WESTT constructs <strong>the</strong> OSD based upon an input <strong>of</strong> <strong>the</strong> HTA for <strong>the</strong><br />
scenario under analysis.<br />
Example<br />
The following example is an extract <strong>of</strong> an OSD <strong>of</strong> an energy distribution scenario<br />
(Salmon et al 2004b). The task involved <strong>the</strong> switching out <strong>of</strong> three circuits (SGT5 and<br />
SGT1A and 1B) at Barking 275Kv, 132Kv and 33Kv Substations. Circuit SGT5 was<br />
being switched out for <strong>the</strong> installation <strong>of</strong> a new transformer for <strong>the</strong> nearby channel tunnel<br />
rail link and SGT1A and 1B were being switched out for substation maintenance.<br />
63
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Observational data from Barking substation and <strong>the</strong> NOC control room was used to<br />
conduct a HTA <strong>of</strong> <strong>the</strong> switching scenario. A HTA was <strong>the</strong>n created, which acted as <strong>the</strong><br />
primary input into <strong>the</strong> OSD diagram. Total co-ordination values for each teamwork task<br />
step are also annotated onto <strong>the</strong> OSD. The glossary for <strong>the</strong> OSD is presented in Figure<br />
3-16. An extract <strong>of</strong> <strong>the</strong> HTA for <strong>the</strong> corresponding energy distribution task is presented<br />
after Figure 3-16. The corresponding extract <strong>of</strong> <strong>the</strong> OSD is presented in Figure 3-17. The<br />
operational loading figures are presented in Table 24.<br />
Table 24 - Operational loading results<br />
Agent Operatio<br />
n<br />
Receive Transport Decision Delay Total<br />
NOC 98 40 138<br />
SAP 223 21 19 1 264<br />
WOK 40 10 50<br />
REC 15 14 29<br />
The operational loading analysis indicates that <strong>the</strong> senior authorised person (SAP) at<br />
Barking substation has <strong>the</strong> highest loading in terms <strong>of</strong> operations, transport, and delay<br />
whilst <strong>the</strong> network operations centre (NOC) operator has <strong>the</strong> highest loading in terms <strong>of</strong><br />
receipt <strong>of</strong> information. This provides an indication <strong>of</strong> <strong>the</strong> nature <strong>of</strong> <strong>the</strong> roles involved in<br />
<strong>the</strong> scenario. The NOC operators role is one <strong>of</strong> information distribution (giving and<br />
receiving) indicated by <strong>the</strong> high receive operator loading, whilst <strong>the</strong> majority <strong>of</strong> <strong>the</strong> work<br />
is conducted by <strong>the</strong> SAP at Barking, indicated by <strong>the</strong> high operation and transport loading<br />
figures.<br />
64
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Figure 3-16 - OSD Glossary (Source: Salmon et al 2004)<br />
0. Co-ordinate and carry out switching operations on circuits SGT5. SGT1A and 1B at<br />
Bark s/s (Plan 0. Do 1 <strong>the</strong>n 2 <strong>the</strong>n 3, EXIT)<br />
1. Prepare for switching operations (Plan 1. Do 1.1, <strong>the</strong>n 1.2, <strong>the</strong>n 1.3, <strong>the</strong>n 1.4, <strong>the</strong>n 1.5,<br />
<strong>the</strong>n 1.6, <strong>the</strong>n 1.7, <strong>the</strong>n 1.8, <strong>the</strong>n 1.9,<strong>the</strong>n 1.10 EXIT)<br />
1.1. Agree SSC (Plan 1.1. Do 1.1.1, <strong>the</strong>n 1.1.2, <strong>the</strong>n 1.1.3, <strong>the</strong>n 1.1.4, <strong>the</strong>n 1.1.5, EXIT)<br />
1.1.1. (WOK) Use phone to Contact NOC<br />
1.1.2. (WOK + NOC) Exchange identities<br />
1.1.3. (WOK + NOC) Agree SSC documentation<br />
1.1.4. (WOK+NOC) Agree SSC and time (Plan 1.1.4. Do 1.1.4.1, <strong>the</strong>n 1.1.4.2,<br />
EXIT)<br />
1.1.4.1. (NOC) Agree SSC with WOK<br />
1.1.4.2. (NOC) Agree time with WOK<br />
1.1.5. (NOC) Record and enter details (Plan 1.1.5. Do 1.1.5.1, <strong>the</strong>n 1.1.5.2, EXIT)<br />
1.1.5.1. Record details on log sheet<br />
65
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
1.1.5.2. Enter details into worksafe<br />
1.2. (NOC) Request remote isolation (Plan 1.2. Do 1.2.1, <strong>the</strong>n 1.2.2, <strong>the</strong>n 1.2.3, <strong>the</strong>n<br />
1.2.4, EXIT)<br />
1.2.1. (NOC) Ask WOK for isolators to be opened remotely<br />
1.2.2. (WOK) Perform remote isolation<br />
1.2.3. (NOC) Check Barking s/s screen<br />
1.2.4. (WOK + NOC) End communications<br />
1.3. Ga<strong>the</strong>r information on outage at transformer 5 at Bark s/s<br />
(Plan 1.3. Do 1.3.1, <strong>the</strong>n 1.3.2, <strong>the</strong>n 1.3.3, <strong>the</strong>n 1.3.4, EXIT)<br />
1.3.1. (NOC) Use phone to contact SAP at Bark<br />
Figure 3-17 - Extract for NGT Switching scenario (Source: Salmon et al 2004)<br />
66
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Flowchart<br />
START<br />
Define task(s) under<br />
analysis<br />
Collect specific data<br />
regarding <strong>the</strong> task<br />
under analysis<br />
Conduct a HTA for <strong>the</strong><br />
task under analysis<br />
Construct Operation<br />
sequence diagram<br />
Calculate operational<br />
loading figures for each<br />
agent involved in <strong>the</strong><br />
task<br />
Add additional<br />
analysis results e.g.<br />
CDA, Comms usage<br />
diagram<br />
STOP<br />
67
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
3.8 Critical Decision Method<br />
Background and applications<br />
The Critical Decision Method is a semi-structured interview technique that uses cognitive<br />
probes in order to elicit information regarding expert decision-making. According to <strong>the</strong><br />
authors, <strong>the</strong> technique can serve to provide knowledge engineering for expert system<br />
development, identify training requirements, generate training materials and evaluate <strong>the</strong><br />
task performance impact <strong>of</strong> expert systems (Klein, Calderwood & MacGregor 1989).<br />
The technique is an extension <strong>of</strong> <strong>the</strong> Critical Incident Technique (Flanagan 1954) and<br />
was developed in order to study <strong>the</strong> naturalistic decision-making strategies <strong>of</strong><br />
experienced personnel. The CDM is perhaps <strong>the</strong> most commonly used cognitive task<br />
analysis technique and has been applied in a number <strong>of</strong> domains, including <strong>the</strong> fire<br />
service (Baber et al 2004), military and paramedics (Klein, Calderwood & MacGregor<br />
1989), air traffic control (Walker et al 2004), civil energy distribution (Salmon et al<br />
2004), naval warfare (Stewart et al 2004), rail (Walker et al 2004) and even white water<br />
rafting (O’Hare et al 2000).<br />
Domain <strong>of</strong> application<br />
Generic.<br />
Procedure and advice<br />
Step 1: Select <strong>the</strong> Incident to be analysed<br />
The first part <strong>of</strong> a CDM analysis is to select <strong>the</strong> incident that is to be analysed.<br />
Depending upon <strong>the</strong> purpose <strong>of</strong> <strong>the</strong> analysis, <strong>the</strong> type <strong>of</strong> incident may already be selected.<br />
CDM normally focuses on non-routine incidents, such as emergency incidents, or highly<br />
challenging incidents. If <strong>the</strong> scenario under analysis is not already specified, <strong>the</strong><br />
analyst(s) may identify an appropriate incident via interview with an appropriate SME,<br />
by asking <strong>the</strong>m to describe a recent highly challenging (i.e. high workload) or non routine<br />
incident in which <strong>the</strong>y were involved. The interviewee involved in <strong>the</strong> CDM analysis<br />
should be <strong>the</strong> primary decision maker in <strong>the</strong> chosen incident.<br />
Step 2: Select CDM probes<br />
The CDM technique works by probing a SME using specific probes designed to elicit<br />
pertinent information regarding <strong>the</strong> decision making process during key points in <strong>the</strong><br />
incident under analysis. In order to ensure that <strong>the</strong> output is compliant with <strong>the</strong> original<br />
aims <strong>of</strong> <strong>the</strong> analysis, an appropriate set <strong>of</strong> CDM probes should be defined prior to <strong>the</strong><br />
analysis. The probes used are dependent upon <strong>the</strong> aims <strong>of</strong> <strong>the</strong> analysis and <strong>the</strong> domain in<br />
which <strong>the</strong> incident is embedded. Alternatively, if <strong>the</strong>re are no adequate probes available,<br />
<strong>the</strong> analyst(s) can develop novel probes based upon <strong>the</strong> analysis needs. A set <strong>of</strong> CDM<br />
probes used in work package 1 to analyse C4i activity are presented in table 9 <strong>of</strong> this<br />
report. Step 3: Select appropriate participant<br />
68
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Once <strong>the</strong> scenario under analysis and <strong>the</strong> probes to be used are defined, <strong>the</strong> analyst(s)<br />
should proceed to identify an appropriate SME or set <strong>of</strong> SME’s. Typically, operators <strong>of</strong><br />
<strong>the</strong> system under analysis are used.<br />
Step 4: Ga<strong>the</strong>r and record account <strong>of</strong> <strong>the</strong> incident<br />
The CDM procedure can be applied to an incident observed by <strong>the</strong> analyst or to a<br />
retrospective incident described by <strong>the</strong> participant. If <strong>the</strong> CDM is based upon an observed<br />
incident, <strong>the</strong>n this step involves firstly observing <strong>the</strong> incident and <strong>the</strong>n recording an<br />
account <strong>of</strong> <strong>the</strong> incident. O<strong>the</strong>rwise, <strong>the</strong> incident can be described retrospectively form <strong>the</strong><br />
participants memory. The analyst should ask <strong>the</strong> SME for a description <strong>of</strong> <strong>the</strong> incident in<br />
question, from its starting point to its end point.<br />
Step 5: Construct Incident Timeline<br />
The next step in <strong>the</strong> CDM analysis is to construct a timeline <strong>of</strong> <strong>the</strong> incident described in<br />
step 4. The aim <strong>of</strong> this is to give <strong>the</strong> analyst(s) a clear picture <strong>of</strong> <strong>the</strong> incident and its<br />
associated events, including when each event occurred and what <strong>the</strong> duration <strong>of</strong> each<br />
event was. According to Klein, Calderwood & MacGregor (1989) <strong>the</strong> events included in<br />
<strong>the</strong> timeline should encompass any physical events, such as alarms sounding, and also<br />
‘mental’ events, such as <strong>the</strong> thoughts and perceptions <strong>of</strong> <strong>the</strong> interviewee during <strong>the</strong><br />
incident.<br />
Step 6: Define scenario phases<br />
Once <strong>the</strong> analyst has a clear understanding <strong>of</strong> <strong>the</strong> incident under analysis, <strong>the</strong> incident<br />
should be divided into key phases or decision points. This should be done in conjunction<br />
with <strong>the</strong> SME. Normally, <strong>the</strong> incident is divided into four or five key phases.<br />
Step 7: Use CDM probes to query participant decision making<br />
For each incident phase, <strong>the</strong> analyst should probe <strong>the</strong> SME using <strong>the</strong> CDM probes<br />
selected during step 2 <strong>of</strong> <strong>the</strong> procedure. The probes are used in an unstructured interview<br />
format in order to ga<strong>the</strong>r pertinent information regarding <strong>the</strong> SME’s decision making<br />
during each incident phase. The interview should be recorded using an audio recording<br />
device such as a mini-disc recorder.<br />
Step 8: Transcribe interview data<br />
Once <strong>the</strong> interview is complete, <strong>the</strong> data should be transcribed accordingly.<br />
Step 9: Construct CDM tables<br />
Finally, a CDM output table for each scenario phase should be constructed. This involves<br />
simply presenting <strong>the</strong> CDM probes and <strong>the</strong> associated SME answers in an output table<br />
(see Table 25 to Table 28).<br />
69
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Advantages<br />
Disadvantages<br />
• The CDM analysis can be used to elicit specific information regarding <strong>the</strong><br />
decision making strategies used in complex systems.<br />
• The CDM output can be used to construct propositional networks which<br />
describe <strong>the</strong> knowledge or SA objects required during <strong>the</strong> scenario under<br />
analysis.<br />
• The technique is normally quick in application.<br />
• Once familiar with <strong>the</strong> technique, CDM is relatively easy to apply.<br />
• The CDM is a popular procedure and has been applied in a number <strong>of</strong><br />
domains.<br />
• The reliability <strong>of</strong> such a technique is questionable. Klein & Armstrong<br />
(2004) suggests that methods that analyse retrospective incidents are<br />
associated with concerns <strong>of</strong> data reliability, due to evidence <strong>of</strong> memory<br />
degradation.<br />
• The data is obtained is highly dependent upon <strong>the</strong> skill <strong>of</strong> <strong>the</strong> analyst<br />
conducting <strong>the</strong> CDM interview and also <strong>the</strong> quality <strong>of</strong> <strong>the</strong> participant used.<br />
• A high level <strong>of</strong> expertise and training is required in order to use <strong>the</strong> CDM to<br />
its maximum effect (Klein & Armstrong 2004).<br />
• The CDM relies upon interviewee verbal reports in order to reconstruct<br />
incidents. How far a verbal report accurately represents <strong>the</strong> cognitive<br />
processes <strong>of</strong> <strong>the</strong> decision maker is questionable. Facts could be easily<br />
misrepresented by <strong>the</strong> participants involved.<br />
• It is <strong>of</strong>ten difficult to gain sufficient access to appropriate SME’s in order to<br />
conduct a CDM analysis.<br />
70
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Flowchart<br />
START<br />
Define scenario(s)<br />
under analysis<br />
Select appropriate<br />
CDM probes<br />
Select appropriate<br />
SME(s)<br />
Ask SME to give a<br />
description <strong>of</strong> incident<br />
Divide <strong>the</strong> incident<br />
into key phases<br />
Take <strong>the</strong> first/next<br />
incident phase<br />
Use probes to ga<strong>the</strong>r<br />
information for each<br />
incident phase<br />
Are <strong>the</strong>re any<br />
more incident<br />
phases<br />
Transcribe data<br />
Construct CDM<br />
output tables<br />
STOP<br />
71
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Example<br />
The following example is taken from a CDM analysis that was conducted in order to<br />
analyse C4i activity in <strong>the</strong> civil energy distribution domain (Salmon et al 2004b). The<br />
scenario under analysis involved <strong>the</strong> switching out <strong>of</strong> three circuits at Barking 275Kv,<br />
132Kv and 33Kv Substations. Circuit SGT5 was being switched out for <strong>the</strong> installation<br />
<strong>of</strong> a new transformer for <strong>the</strong> nearby channel tunnel rail link and SGT1A and 1B were<br />
being switched out for substation maintenance. For <strong>the</strong> CDM analysis, <strong>the</strong> control room<br />
operator co-ordinating <strong>the</strong> activity and <strong>the</strong> senior authorised person at Barking substation<br />
who conducted <strong>the</strong> activity were interviewed. The set <strong>of</strong> CDM probes used are presented<br />
in Table 9. The scenario was divided into four key phases:<br />
First issue <strong>of</strong> instructions<br />
• Deal with switching requests<br />
• Perform isolation<br />
• Report back to network operations centre<br />
The CDM output is presented in Table 25 to Table 28.<br />
Table 25 - Phase 1: First issue <strong>of</strong> instructions<br />
Goal<br />
Specification<br />
Cue<br />
identification<br />
Expectancy<br />
Conceptual<br />
Model<br />
Uncertainty<br />
Information<br />
Situation<br />
Awareness<br />
Situation<br />
Assessment<br />
Options<br />
Stress<br />
Choice<br />
Analogy<br />
Establish what isolation <strong>the</strong> SAP at Barking is looking for. Depends on gear<br />
Don’t Believe It (DBI) alarm is unusual – faulty contact (not open or closed)<br />
questionable data from site checking rating <strong>of</strong> earth switches (may be not fully rated<br />
for circuit current – so additional earths may be required.<br />
Check that SAP is happy with instructions as not normal.<br />
Decision expected by DBI is not common.<br />
Recognised instruction but not stated in WE1000 – as <strong>the</strong>re are not too many front<br />
and rear shutters metal clad switch gear.<br />
Confirm from field about planned instruction – make sure that SAP is happy with <strong>the</strong><br />
instruction.<br />
Reference to front and rear busbars.<br />
WE1000 procedure<br />
Metal clad switchgear<br />
Barking SGT1A/1B substation screen<br />
SAP at Barking<br />
Ask colleagues if needed to<br />
No alternatives<br />
N/A<br />
WE1000 – need to remove what does not apply<br />
Could add front and rear busbar procedures<br />
Best practice guide for metal clad EMS switching<br />
72
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 26 - Phase 2: Deal with switching requests<br />
Goal<br />
Specification<br />
Cue<br />
identification<br />
Expectancy<br />
Conceptual<br />
Model<br />
Uncertainty<br />
Information<br />
Situation<br />
Awareness<br />
Situation<br />
Assessment<br />
Options<br />
Stress<br />
Choice<br />
Analogy<br />
Obtain confirmation from NOC that planned isolation is still required.<br />
Approaching time for planned isolation.<br />
Switching phone rings throughout building.<br />
Airblast circuit breakers (accompanied by sirens) can be heard to operate remotely<br />
(more so in Barking 275 than Barking C 132).<br />
Yes – routine planned work according to fixed procedures.<br />
Wokingham have performed remote isolations already.<br />
Circuit configured ready for local isolation.<br />
Physical verification <strong>of</strong> apparatus always required (DBI – don’t believe it).<br />
Proceduralised information from NOC – circuit, location, time, actions required etc.<br />
Switching log.<br />
Switching log.<br />
Physical status <strong>of</strong> apparatus.<br />
Planning documentation.<br />
Visual or verbal information from substation personnel.<br />
Planning documentation used only occasionally<br />
Refusal <strong>of</strong> switching request.<br />
Additional conditions to switching request.<br />
Some time pressure.<br />
Yes – highly proceduralised anyway<br />
Yes – routine activity<br />
Table 27 - Phase 3: Perform Isolation<br />
Goal<br />
Specification<br />
Cue<br />
identification<br />
Expectancy<br />
Conceptual<br />
Model<br />
Uncertainty<br />
Information<br />
Situation<br />
Awareness<br />
Situation<br />
Assessment<br />
Options<br />
Stress<br />
Choice<br />
Analogy<br />
Inform NOC <strong>of</strong> isolation status<br />
Switching telephone.<br />
NOC operator answers<br />
NOC accepts.<br />
Manner in which circuit is now isolated.<br />
Form <strong>of</strong> procedures.<br />
No – possibly fur<strong>the</strong>r instructions, possibly mismatches local situation and remote<br />
displays in NOC.<br />
Switching log.<br />
Verbal information from NOC.<br />
Switching log.<br />
Yes – all information used.<br />
No (raise or add on fur<strong>the</strong>r requests etc. to <strong>the</strong> same call)<br />
No<br />
Yes – highly proceduralised<br />
Yes – frequently performed activity<br />
73
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 28 - Phase 4: Report back to NOC<br />
Goal<br />
Specification<br />
Cue<br />
identification<br />
Expectancy<br />
Conceptual<br />
Model<br />
Uncertainty<br />
Information<br />
Situation<br />
Awareness<br />
Situation<br />
Assessment<br />
Options<br />
Stress<br />
Choice<br />
Analogy<br />
Ensure it is safe to perform local isolation.<br />
Confirm circuits/equipment to be operated.<br />
Telecontrol displays/circuit loadings.<br />
Equipment labels.<br />
Equipment displays.<br />
O<strong>the</strong>r temporary notices.<br />
Equipment configured according to planned circuit switching.<br />
Equipment will function correctly.<br />
Layout/type/characteristics <strong>of</strong> circuit.<br />
Circuit loadings/balance.<br />
Function <strong>of</strong> equipment.<br />
Will equipment physically work as expected (will something jam etc.).<br />
O<strong>the</strong>r work being carried out by o<strong>the</strong>r parties (e.g. EDF).<br />
Switching log.<br />
Visual and verbal information from those undertaking <strong>the</strong> work.<br />
Physical information from apparatus and telecontrol displays.<br />
All information used<br />
Inform NOC that isolation cannot be performed/o<strong>the</strong>r aspects <strong>of</strong> switching<br />
instructions cannot be carried out.<br />
Some time pressure.<br />
Possibly some difficulties in operating or physically handling <strong>the</strong> equipment.<br />
Yes – proceduralised within equipment types. Occasional non-routine activities<br />
required to cope with unusual/unfamiliar equipment, or equipment not owned by<br />
NGT.<br />
Yes – <strong>of</strong>ten. Except in cases with unfamiliar equipment.<br />
Related Methods<br />
The CDM is an extension <strong>of</strong> <strong>the</strong> critical incident technique (Flanagan 1954). The CDM<br />
is also closely related to o<strong>the</strong>r cognitive task analysis (CTA) techniques, in that it uses<br />
probes to elicit data regarding task performance from participants. O<strong>the</strong>r similar CTA<br />
techniques include ACTA (Militello and Hutton 2000) and cognitive walkthrough<br />
analysis (Polson et al 1992).<br />
Approximate training and application times<br />
Klein & Armstrong (2004) report that <strong>the</strong> training time associated with <strong>the</strong> CDM would<br />
be high. Experience in interviews with SME’s is required, and also a grasp <strong>of</strong> cognitive<br />
psychology. The application time for <strong>the</strong> CDM is medium. The CDM interview takes<br />
between 1 - 2 hours, and <strong>the</strong> transcription process takes approximately 1 – 2 hours.<br />
Reliability and validity<br />
Both intra and inter analyst reliability <strong>of</strong> <strong>the</strong> CDM approach is questionable. It is<br />
apparent that such an approach may elicit different data from similar incidents when<br />
applied by different analysts on separate participants. Klein & Armstrong (2004)<br />
suggests that <strong>the</strong>re are also concerns associated with <strong>the</strong> reliability <strong>of</strong> <strong>the</strong> CDM due to<br />
evidence <strong>of</strong> memory degradation.<br />
74
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Tools needed<br />
When conducting a CDM analysis, pen and paper could be sufficient. However, to<br />
ensure that data collection is comprehensive, it is recommended that video or audio<br />
recording equipment is used. A set <strong>of</strong> CDM probes, such as those presented in Table 9<br />
are also required. The type <strong>of</strong> probes used is dependent upon <strong>the</strong> focus <strong>of</strong> <strong>the</strong> analysis.<br />
3.9 Propositional Networks<br />
Background and applications<br />
Propositional networks are used identify <strong>the</strong> knowledge objects related to a particular task<br />
or scenario, and also <strong>the</strong> links between each <strong>of</strong> <strong>the</strong> knowledge objects identified.<br />
According to Baber and Stanton (2004), <strong>the</strong> concept <strong>of</strong> representing ‘knowledge’ in <strong>the</strong><br />
form <strong>of</strong> a network has been subjected to major discussion within cognitive psychology<br />
since <strong>the</strong> 1970s. Propositional networks consist <strong>of</strong> a set <strong>of</strong> nodes that represent<br />
knowledge, sources <strong>of</strong> information, agents, and artefacts that are linked through specific<br />
causal paths. Thus <strong>the</strong> propositional network <strong>of</strong>fers a way <strong>of</strong> presenting <strong>the</strong> ‘ideal’<br />
collection <strong>of</strong> knowledge required during <strong>the</strong> scenario in question. Networks are<br />
constructed from an initial critical decision method analysis <strong>of</strong> <strong>the</strong> scenario in question. A<br />
simple content analysis is used to identify <strong>the</strong> knowledge objects for each scenario phase<br />
as identified by <strong>the</strong> CDM analysis. A propositional network is <strong>the</strong>n constructed for each<br />
phase identified by <strong>the</strong> CDM analysis, comprised <strong>of</strong> <strong>the</strong> knowledge objects and <strong>the</strong> links<br />
between <strong>the</strong>m. Propositional networks have been used to represent knowledge and shared<br />
situation awareness as part <strong>of</strong> <strong>the</strong> EAST methodology (Baber et al 2004), which has been<br />
used to analyse C4i scenarios in <strong>the</strong> civil energy distribution domain (Salmon et al 2004b<br />
+ c), <strong>the</strong> rail domain (Walker et al 2004), <strong>the</strong> air traffic control domain (Walker et al<br />
2004), <strong>the</strong> naval warfare domain (Stewart et al 2004) and <strong>the</strong> emergency services domain<br />
(Baber et al 2004).<br />
Domain <strong>of</strong> application<br />
Generic.<br />
Procedure and Advice<br />
Step 1: Define scenario<br />
The first step in a propositional network analysis is to define <strong>the</strong> scenario under analysis.<br />
The scenario in question should be defined clearly. This allows <strong>the</strong> analyst(s) to<br />
determine <strong>the</strong> data collection procedure that follows and also <strong>the</strong> appropriate SME’s<br />
required for <strong>the</strong> CDM phase <strong>of</strong> <strong>the</strong> analysis.<br />
Step 2: Conduct a HTA for <strong>the</strong> scenario<br />
Once <strong>the</strong> scenario has been clearly defined, <strong>the</strong> next step involves describing <strong>the</strong> scenario<br />
using HTA. A number <strong>of</strong> data collection techniques may be used in order to ga<strong>the</strong>r <strong>the</strong><br />
75
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
information required for <strong>the</strong> HTA, such as interviews with SME’s and observations <strong>of</strong> <strong>the</strong><br />
task under analysis.<br />
Step 3: Conduct a CDM analysis<br />
The propositional networks are based upon a CDM analysis <strong>of</strong> <strong>the</strong> scenario in question.<br />
The CDM analysis should be conducted using appropriate SME’s (see section 3.8 for a<br />
full description <strong>of</strong> <strong>the</strong> CDM procedure). The CDM involves dividing <strong>the</strong> scenario under<br />
analysis into a number <strong>of</strong> key phases and <strong>the</strong>n probing <strong>the</strong> SME using pre-defined<br />
‘cognitive’ probes, designed to determine pertinent features associated with decision<br />
making during each scenario phase.<br />
Step 4: Conduct content analysis<br />
Once <strong>the</strong> CDM data is collected, a simple content analysis should be conducted for each<br />
phase identified during <strong>the</strong> CDM analysis. In order to convert <strong>the</strong> CDM tables into<br />
propositions, a content analysis is performed. In <strong>the</strong> first stage, this simply means<br />
separating all content words from any function words. For example, <strong>the</strong> entry in table one<br />
“Respiratory problems caused by unknown, airborne material” would be reduced to <strong>the</strong><br />
following propositions ‘respiratory problems’, ‘airborne’ and ‘material’. Working<br />
through <strong>the</strong> table leads to a set <strong>of</strong> propositions. These are checked to ensure that<br />
duplication is minimised and <strong>the</strong>n used to construct <strong>the</strong> propositional network.<br />
In order to specify <strong>the</strong> knowledge objects for each phase, <strong>the</strong> analyst simply takes <strong>the</strong><br />
CDM output for each phase and using a simple content analysis, identifies <strong>the</strong> required<br />
knowledge objects. Knowledge objects include any knowledge, information, agents and<br />
artefacts identified by <strong>the</strong> CDM analysis. A simple list <strong>of</strong> knowledge objects should be<br />
made for each scenario phase.<br />
Step 5: Define links between knowledge objects<br />
Once <strong>the</strong> knowledge objects for each scenario phase have been identified, <strong>the</strong> next step<br />
involves defining <strong>the</strong> links between <strong>the</strong> knowledge objects in each phase. The following<br />
knowledge objects links taxonomy is used:<br />
• Has<br />
• Is<br />
• Causes<br />
• Knows<br />
• Requires<br />
• Prevents<br />
76
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
For those knowledge objects that are linked during <strong>the</strong> scenario, <strong>the</strong> type <strong>of</strong> link should<br />
be defined using <strong>the</strong> links taxonomy above.<br />
Step 6: Construct propositional networks<br />
The final step is to construct <strong>the</strong> propositional network diagrams for each scenario phase.<br />
A propositional network diagram should be constructed for <strong>the</strong> overall scenario (i.e.<br />
including all knowledge objects) and <strong>the</strong>n separate propositional network diagrams<br />
should be constructed for each phase, with <strong>the</strong> knowledge objects required highlighted in<br />
red. Fur<strong>the</strong>r coding <strong>of</strong> <strong>the</strong> knowledge objects may also be used e.g. shared knowledge<br />
objects can be striped in colour, and inactive knowledge objects that have been used in<br />
previous scenario phases are typically shaded.<br />
Advantages<br />
Disadvantages<br />
• The output represents <strong>the</strong> ideal collection <strong>of</strong> knowledge required for<br />
performance during <strong>the</strong> scenario under analysis.<br />
• The knowledge objects are defined for each phase <strong>of</strong> <strong>the</strong> scenario under<br />
analysis, and <strong>the</strong> links between <strong>the</strong> knowledge objects are also specified.<br />
• The technique is easy to learn and use.<br />
• The technique is also quick in its application.<br />
• Propositional networks are ideal for analysing teamwork and representing<br />
shared situation awareness during a particular scenario.<br />
• The initial HTA and CDM analysis add considerable time to <strong>the</strong> associated<br />
application time.<br />
• Inter and intra analyst reliability <strong>of</strong> <strong>the</strong> technique is questionable.<br />
• A propositional network analysis is reliant upon acceptable CDM data.<br />
• It may be difficult to ga<strong>the</strong>r appropriate SME’s for <strong>the</strong> CDM part <strong>of</strong> <strong>the</strong><br />
analysis.<br />
77
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Flowchart<br />
START<br />
Define scenario<br />
under analysis<br />
Conduct a HTA for<br />
<strong>the</strong> scenario under<br />
analysis<br />
Conduct a CDM<br />
analysis with SME(s)<br />
Take first/next CDM<br />
phase<br />
Identify knowledge<br />
objects using content<br />
analysis<br />
Define knowledge object<br />
links using <strong>the</strong> associated<br />
links taxonomy<br />
Construct propositional<br />
network<br />
Are <strong>the</strong>re any<br />
more CDM<br />
phases<br />
STOP<br />
78
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Example<br />
The following example is taken from an analysis <strong>of</strong> a switching scenario drawn from <strong>the</strong><br />
civil energy distribution domain (Salmon et al 2004b). The propositional networks<br />
presented in Figure 3-18 to Figure 3-22 present <strong>the</strong> knowledge objects (shaded in red)<br />
identified from <strong>the</strong> corresponding CDM output for that phase. The CDM outputs are<br />
presented in Table 29 to Table 32. The propositional network consists <strong>of</strong> a set <strong>of</strong> nodes<br />
that represent sources <strong>of</strong> information, agents, and objects etc. that are linked through<br />
specific causal paths. From this network, it is possible to identify required information<br />
and possible options relevant to this incident. The concept behind using a propositional<br />
network in this manner is that it represents <strong>the</strong> ‘ideal’ collection <strong>of</strong> knowledge for <strong>the</strong><br />
scenario. As <strong>the</strong> incident unfolds, so participants will have access to more <strong>of</strong> this<br />
knowledge (ei<strong>the</strong>r through communication with o<strong>the</strong>r agents or through recognising<br />
changes in <strong>the</strong> incident status). Consequently, within this propositional network, Situation<br />
Awareness can be represented as <strong>the</strong> change in weighting <strong>of</strong> links. Propositional<br />
networks were developed for <strong>the</strong> overall scenario and also <strong>the</strong> incident phases identified<br />
during <strong>the</strong> CDM analysis. The propositional networks indicate which <strong>of</strong> <strong>the</strong> knowledge<br />
objects are active (i.e. agents are using <strong>the</strong>m) during each incident phase. The light blue<br />
nodes in <strong>the</strong> propositional networks represent unactivated knowledge objects (i.e.<br />
knowledge is available but is not required nor is it being used). The red nodes represent<br />
active (or currently being used) knowledge objects.<br />
Table 29 - CDM Phase 1: First issue <strong>of</strong> instructions<br />
Goal Specification<br />
Cue identification<br />
Establish what isolation <strong>the</strong> SAP at Barking is looking for. Depends on gear<br />
Don’t Believe It (DBI) alarm is unusual – faulty contact (not open or closed)<br />
questionable data from site checking rating <strong>of</strong> earth switches (may be not fully<br />
rated for circuit current – so additional earths may be required.<br />
Check that SAP is happy with instructions as not normal.<br />
Expectancy<br />
Conceptual Model<br />
Uncertainty<br />
Information<br />
Situation<br />
Awareness<br />
Decision expected by DBI is not common.<br />
Recognised instruction but not stated in WE1000 – as <strong>the</strong>re are not too many front<br />
and rear shutters metal clad switch gear.<br />
Confirm from field about planned instruction – make sure that SAP is happy with<br />
<strong>the</strong> instruction.<br />
Reference to front and rear busbars.<br />
WE1000 procedure<br />
Metal clad switchgear<br />
Barking SGT1A/1B substation screen<br />
SAP at Barking<br />
Situation<br />
Assessment<br />
Ask colleagues if needed to<br />
79
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Options<br />
Stress<br />
Choice<br />
No alternatives<br />
N/A<br />
WE1000 – need to remove what does not apply<br />
Could add front and rear busbar procedures<br />
Analogy<br />
Best practice guide for metal clad EMS switching<br />
Table 30 - CDM Phase 2: Deal with switching requests<br />
Goal Specification<br />
Cue identification<br />
Expectancy<br />
Conceptual Model<br />
Uncertainty<br />
Information<br />
Situation<br />
Awareness<br />
Situation<br />
Assessment<br />
Options<br />
Stress<br />
Choice<br />
Analogy<br />
Obtain confirmation from NOC that planned isolation is still required.<br />
Approaching time for planned isolation.<br />
Switching phone rings throughout building.<br />
Airblast circuit breakers (accompanied by sirens) can be heard to operate remotely<br />
(more so in Barking 275 than Barking C 132).<br />
Yes – routine planned work according to fixed procedures.<br />
Wokingham have performed remote isolations already.<br />
Circuit configured ready for local isolation.<br />
Physical verification <strong>of</strong> apparatus always required (DBI – don’t believe it).<br />
Proceduralised information from NOC – circuit, location, time, actions required etc.<br />
Switching log.<br />
Switching log.<br />
Physical status <strong>of</strong> apparatus.<br />
Planning documentation.<br />
Visual or verbal information from substation personnel.<br />
Planning documentation used only occasionally<br />
Refusal <strong>of</strong> switching request.<br />
Additional conditions to switching request.<br />
Some time pressure.<br />
Yes – highly proceduralised anyway<br />
Yes – routine activity<br />
Table 31 – CDM Phase 3: Perform isolation<br />
80
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Goal Specification<br />
Cue identification<br />
Expectancy<br />
Conceptual Model<br />
Uncertainty<br />
Information<br />
Situation Awareness<br />
Situation Assessment<br />
Options<br />
Stress<br />
Choice<br />
Analogy<br />
Ensure it is safe to perform local isolation.<br />
Confirm circuits/equipment to be operated.<br />
Telecontrol displays/circuit loadings.<br />
Equipment labels.<br />
Equipment displays.<br />
O<strong>the</strong>r temporary notices.<br />
Equipment configured according to planned circuit switching.<br />
Equipment will function correctly.<br />
Layout/type/characteristics <strong>of</strong> circuit.<br />
Circuit loadings/balance.<br />
Function <strong>of</strong> equipment.<br />
Will equipment physically work as expected (will something jam etc.).<br />
O<strong>the</strong>r work being carried out by o<strong>the</strong>r parties (e.g. EDF).<br />
Switching log.<br />
Visual and verbal information from those undertaking <strong>the</strong> work.<br />
Physical information from apparatus and telecontrol displays.<br />
All information used<br />
Inform NOC that isolation cannot be performed/o<strong>the</strong>r aspects <strong>of</strong> switching instructions<br />
cannot be carried out.<br />
Some time pressure.<br />
Possibly some difficulties in operating or physically handling <strong>the</strong> equipment.<br />
Yes – proceduralised within equipment types. Occasional non-routine activities required<br />
to cope with unusual/unfamiliar equipment, or equipment not owned by NGT.<br />
Yes – <strong>of</strong>ten. Except in cases with unfamiliar equipment.<br />
81
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Table 32 - Phase 4: Report back to NOC<br />
Goal Specification<br />
Cue identification<br />
Expectancy<br />
Conceptual Model<br />
Uncertainty<br />
Information<br />
Situation<br />
Awareness<br />
Situation<br />
Assessment<br />
Options<br />
Stress<br />
Choice<br />
Analogy<br />
Inform NOC <strong>of</strong> isolation status.<br />
Switching telephone.<br />
NOC operator answers.<br />
NOC accepts.<br />
Manner in which circuit is now isolated.<br />
Form <strong>of</strong> procedures.<br />
No – possibly fur<strong>the</strong>r instructions, possibly mismatches local situation and remote<br />
displays in NOC.<br />
Switching log.<br />
Verbal information from NOC.<br />
Switching log.<br />
Yes – all information used.<br />
No (raise or add on fur<strong>the</strong>r requests etc. to <strong>the</strong> same call)<br />
No<br />
Yes – highly proceduralised<br />
Yes – frequently performed activity<br />
82
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Gas Insulated<br />
Circuit<br />
Breakers<br />
Airblast<br />
Notices/Locks<br />
NOC<br />
Check Open<br />
Lock & Caution<br />
Location<br />
SystemState<br />
Certificates<br />
Isolation<br />
Control<br />
Engineer<br />
Open Lock &<br />
Caution<br />
Accept<br />
Instructions<br />
Switching<br />
Wokingham<br />
Shutters<br />
Time<br />
Dressed<br />
Log Sheet<br />
Procedures<br />
Rear<br />
Refuse<br />
Switching Log<br />
Displays<br />
Front<br />
Earth Switches<br />
Isolators<br />
WE1000<br />
Points <strong>of</strong><br />
Isolation<br />
Open<br />
Busbar<br />
Circuits<br />
Electrical<br />
Contacts<br />
Closed<br />
Phone<br />
Current<br />
Voltage<br />
Faulty<br />
DBI Indication<br />
Switching<br />
Phone<br />
Equipment<br />
Lables<br />
Identity<br />
SAP<br />
Substations<br />
Figure 3-18 - Propositional Network for Objects referred to in CDM tables.<br />
Gas Insulated<br />
Circuit<br />
Breakers<br />
Airblast<br />
Notices/Locks<br />
NOC<br />
Check Open<br />
Lock & Caution<br />
Location<br />
SystemState<br />
Certificates<br />
Isolation<br />
Control<br />
Engineer<br />
Open Lock &<br />
Caution<br />
Accept<br />
Instructions<br />
Switching<br />
Wokingham<br />
Shutters<br />
Time<br />
Dressed<br />
Log Sheet<br />
Procedures<br />
Rear<br />
Refuse<br />
Switching Log<br />
Displays<br />
Front<br />
Earth Switches<br />
Isolators<br />
WE1000<br />
Points <strong>of</strong><br />
Isolation<br />
Open<br />
Busbar<br />
Circuits<br />
Electrical<br />
Contacts<br />
Closed<br />
Phone<br />
Current<br />
Voltage<br />
Faulty<br />
DBI Indication<br />
Switching<br />
Phone<br />
Equipment<br />
Lables<br />
Identity<br />
SAP<br />
Substations<br />
Figure 3-19 - Propositional Network for CDM phase one.<br />
83
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Gas Insulated<br />
Circuit<br />
Breakers<br />
Airblast<br />
Notices/Locks<br />
NOC<br />
Check Open<br />
Lock &Caution<br />
Location<br />
SystemState<br />
Certificates<br />
Isolation<br />
Control<br />
Engineer<br />
Open Lock &<br />
Caution<br />
Accept<br />
Instructions<br />
Switching<br />
Wokingham<br />
Shutters<br />
Time<br />
Dressed<br />
Log Sheet<br />
Procedures<br />
Rear<br />
Refuse<br />
Switching Log<br />
Displays<br />
Front<br />
Earth Switches<br />
Isolators<br />
WE1000<br />
Points <strong>of</strong><br />
Isolation<br />
Open<br />
Busbar<br />
Circuits<br />
Electrical<br />
Contacts<br />
Closed<br />
Phone<br />
Current<br />
Voltage<br />
Faulty<br />
DBI Indication<br />
Switching<br />
Phone<br />
Equipment<br />
Lables<br />
Identity<br />
SAP<br />
Substations<br />
Figure 3-20 - Propositional Network for CDM phase two.<br />
84
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Gas Insulated<br />
Circuit<br />
Breakers<br />
Airblast<br />
Notices/Locks<br />
NOC<br />
Check Open<br />
Lock &Caution<br />
Location<br />
SystemState<br />
Certificates<br />
Isolation<br />
Control<br />
Engineer<br />
Open Lock &<br />
Caution<br />
Accept<br />
Instructions<br />
Switching<br />
Wokingham<br />
Shutters<br />
Time<br />
Dressed<br />
Log Sheet<br />
Procedures<br />
Rear<br />
Refuse<br />
Switching Log<br />
Displays<br />
Front<br />
Earth Switches<br />
Isolators<br />
WE1000<br />
Points <strong>of</strong><br />
Isolation<br />
Open<br />
Busbar<br />
Circuits<br />
Electrical<br />
Contacts<br />
Closed<br />
Phone<br />
Current<br />
Voltage<br />
Faulty<br />
DBI Indication<br />
Switching<br />
Phone<br />
Equipment<br />
Lables<br />
Identity<br />
SAP<br />
Substations<br />
Figure 3-21 - Propositional Network for CDM phase three.<br />
Gas Insulated<br />
Circuit<br />
Breakers<br />
Airblast<br />
Notices/Locks<br />
NOC<br />
Check Open<br />
Lock & Caution<br />
Location<br />
SystemState<br />
Certificates<br />
Isolation<br />
Control<br />
Engineer<br />
Open Lock &<br />
Caution<br />
Accept<br />
Instructions<br />
Switching<br />
Wokingham<br />
Shutters<br />
Time<br />
Dressed<br />
Log Sheet<br />
Procedures<br />
Rear<br />
Refuse<br />
Switching Log<br />
Displays<br />
Front<br />
Earth Switches<br />
Isolators<br />
WE1000<br />
Points <strong>of</strong><br />
Isolation<br />
Open<br />
Busbar<br />
Circuits<br />
Electrical<br />
Contacts<br />
Closed<br />
Phone<br />
Current<br />
Voltage<br />
Faulty<br />
DBI Indication<br />
Switching<br />
Phone<br />
Equipment<br />
Lables<br />
Identity<br />
SAP<br />
Substations<br />
Figure 3-22 - Propositional Network for CDM phase four.<br />
85
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Related methods<br />
Propositional networks require an initial CDM analysis as an input. A HTA is also<br />
typically conducted prior to <strong>the</strong> propositional network analysis. The technique has also<br />
been used in conjunction with a number <strong>of</strong> o<strong>the</strong>r techniques (HTA, observation, coordination<br />
demands analysis, comms usage diagram, social network analysis) in <strong>the</strong> form<br />
<strong>of</strong> <strong>the</strong> event analysis <strong>of</strong> systemic teamwork (EAST) methodology (Baber et al 2004),<br />
which has been used to analyse C4i activity in a number <strong>of</strong> domains.<br />
Approximate training and application times<br />
The propositional network methodology requires only minimal training. In a recent HF<br />
methods training session, <strong>the</strong> training time for <strong>the</strong> propositional network technique was<br />
approximately one hour. However, <strong>the</strong> analyst should be competent in <strong>the</strong> HTA and<br />
CDM procedure in order to conduct <strong>the</strong> analysis properly. The application time for<br />
propositional networks alone is high, as it involves a content analysis (on CDM outputs)<br />
and also <strong>the</strong> construction <strong>of</strong> <strong>the</strong> propositional networks.<br />
Reliability and validity<br />
No data regarding <strong>the</strong> reliability and validity <strong>of</strong> <strong>the</strong> technique are available. From<br />
previous experience, it is evident that <strong>the</strong> reliability <strong>of</strong> <strong>the</strong> technique may be questionable.<br />
Certainly, different analysts may identify different knowledge objects for <strong>the</strong> same<br />
scenario (Intra-analyst reliability). Also, <strong>the</strong> same analyst may identify different<br />
knowledge objects for <strong>the</strong> same scenario on different occasions (Inter-analyst reliability).<br />
Tools needed<br />
A propositional network analysis can be conducted using pen and paper. However, it is<br />
recommended that during <strong>the</strong> CDM procedure, an audio recording device is used. When<br />
constructing <strong>the</strong> propositional network diagrams it is recommended that Micros<strong>of</strong>t Visio<br />
is used.<br />
86
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
4.1 Aim<br />
4 Feedback from Questionnaire<br />
The aim <strong>of</strong> this investigation was to understand <strong>the</strong> user acceptance <strong>of</strong> <strong>the</strong> EAST<br />
methodology by its users.<br />
4.2 Participants<br />
Nine participants responded to <strong>the</strong> questionnaire, 8 Male and 1 Female. All had a<br />
minimum <strong>of</strong> a Masters degree with 5 <strong>of</strong> <strong>the</strong> group having a PhD. The sample was split<br />
over four institutions including: Brunel University; Birmingham University; Cranfield<br />
University; and Lockheed Martin. The method had been applied over a wide range <strong>of</strong><br />
situations including. Air force (E3D), Naval (HMS Dryad), Army (a <strong>the</strong>oretical ‘Scud<br />
Hunt’ paradigm), Air Traffic Control, Rail, Energy Distribution (NGC), and <strong>the</strong><br />
Emergency Services (police / fire).<br />
4.3 Equipment<br />
Questionnaires<br />
4.4 Procedure<br />
The questionnaire was sent out via email to a number <strong>of</strong> users within <strong>the</strong> HFI-DTC<br />
consortium who had used <strong>the</strong> EAST Method. The questionnaire was completed<br />
electronically and returned via email.<br />
The questionnaire captured demographic data in <strong>the</strong> first section. In <strong>the</strong> second section<br />
<strong>the</strong> questionnaire captured participant opinion <strong>of</strong> each <strong>of</strong> <strong>the</strong> methodologies in turn that<br />
make up EAST:<br />
Observation<br />
HTA – Hierarchical Task <strong>Analysis</strong><br />
CDA – Coordination Demand <strong>Analysis</strong><br />
CUD – Comms. Usage Diagram<br />
SNA – Social Network <strong>Analysis</strong><br />
OSD – Operation Sequence Diagrams<br />
CDM – Critical Decision Method<br />
PN – Propositional Networks<br />
87
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Participants were asked to rate each <strong>of</strong> <strong>the</strong> methods against <strong>the</strong> following criteria:<br />
• Ease <strong>of</strong> learning<br />
• Time to learn<br />
• Ease <strong>of</strong> application<br />
• Time to Apply<br />
• Usefulness <strong>of</strong> output<br />
• Predictability <strong>of</strong> Output<br />
• Validity <strong>of</strong> output<br />
The participants were asked to respond on a Likert scale that was rated so that 5 was a<br />
positive reaction and 1 a negative reaction. In <strong>the</strong> third and final section <strong>of</strong> <strong>the</strong><br />
questionnaire participants were asked to evaluate <strong>the</strong> methodology as a whole. (Questions<br />
posed can be seen in <strong>the</strong> results section)<br />
4.5 Results<br />
All <strong>of</strong> <strong>the</strong> respondents stated that <strong>the</strong>y would use <strong>the</strong> East methodology again. When<br />
questioned in more detail about which <strong>of</strong> <strong>the</strong> individual tools <strong>the</strong>y would use again; <strong>the</strong>re<br />
was not 100% acceptance for all <strong>of</strong> <strong>the</strong> methods. The results <strong>of</strong> this can be seen in Figure<br />
4-1 <strong>the</strong> methods that did not receive 100% acceptance were <strong>the</strong> CDA, CUD, and OSD<br />
methods.<br />
88
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Percentage who would use method again<br />
100%<br />
100% 100%<br />
100%<br />
100% 100%<br />
90%<br />
89%<br />
89%<br />
80%<br />
78%<br />
70%<br />
60%<br />
50%<br />
40%<br />
30%<br />
20%<br />
10%<br />
0%<br />
Observation HTA CDA CUD SNA OSD CDM PN<br />
Figure 4-1 - Percentage who would use method again<br />
The results collected are plotted on radar plots. A red line has been added to <strong>the</strong><br />
document so that anything outside <strong>the</strong> line can be considered positive and inside <strong>the</strong> line<br />
negative. The original questionnaire along with fur<strong>the</strong>r plots can be found in <strong>the</strong><br />
Appendix <strong>of</strong> this document.<br />
Observation<br />
Observation scores above average on each <strong>of</strong> <strong>the</strong> criteria with <strong>the</strong> exception <strong>of</strong> <strong>the</strong> time to<br />
apply. This method is rated as <strong>the</strong> easiest to learn and scores very highly on validity and<br />
usefulness.<br />
89
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Observation<br />
Ease <strong>of</strong> learning<br />
Validity <strong>of</strong> output<br />
Time to learn<br />
Predictability <strong>of</strong><br />
Output<br />
Ease <strong>of</strong><br />
application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Figure 4-2 - Radar plot <strong>of</strong> medians for observation<br />
HTA – Hierarchical Task <strong>Analysis</strong><br />
HTA scores very highly on usefulness and validity <strong>of</strong> output. However when looking at<br />
<strong>the</strong> means <strong>the</strong> method performs negatively for <strong>the</strong> remaining attributes which<br />
predominantly relate to <strong>the</strong> ease and time to learn and apply. The predictability <strong>of</strong> <strong>the</strong><br />
output is also brought into question with <strong>the</strong> below average rating.<br />
HTA<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
3<br />
Time to learn<br />
2<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Figure 4-3 - Radar plot <strong>of</strong> medians for HTA<br />
90
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
CDA – Coordination Demand <strong>Analysis</strong><br />
The majority <strong>of</strong> <strong>the</strong> ratings <strong>of</strong> <strong>the</strong> attributes <strong>of</strong> CDA fall on or near <strong>the</strong> neutral line when<br />
looking at <strong>the</strong> means. The method scores higher on <strong>the</strong> ease and time to learn and also<br />
<strong>the</strong> ease <strong>of</strong> application.<br />
CDA<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
3<br />
Time to learn<br />
2<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
CUD – Comms Usage Diagram<br />
Figure 4-4 - Radar plot <strong>of</strong> medians for CDA<br />
The CUD shows similar results to <strong>the</strong> CDA method. This method scores poorly for both<br />
predictability and usefulness. The method scores higher on <strong>the</strong> ease and time to learn and<br />
apply.<br />
91
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
CUD<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
3<br />
Time to learn<br />
2<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
SNA – Social Network <strong>Analysis</strong><br />
Figure 4-5 - Radar plot <strong>of</strong> medians for CUD<br />
The majority <strong>of</strong> <strong>the</strong> ratings <strong>of</strong> <strong>the</strong> attributes <strong>of</strong> SNA fall on or near <strong>the</strong> neutral line when<br />
looking at <strong>the</strong> means. The method scores higher on <strong>the</strong> validity, predictability and<br />
usefulness <strong>of</strong> <strong>the</strong> output.<br />
SNA<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
3<br />
Time to learn<br />
2<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Figure 4-6 - Radar plot <strong>of</strong> medians for SNA<br />
92
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
OSD – Operation Sequence Diagrams<br />
As with <strong>the</strong> SNA and CDM method this method shows fairy indifferent responses to <strong>the</strong><br />
ease and time to apply and learn. It scores highly on validity <strong>of</strong> output. The method<br />
received negative feedback on <strong>the</strong> predictability <strong>of</strong> <strong>the</strong> output.<br />
OSD<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
3<br />
Time to learn<br />
2<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
CDM – Critical Decision Method<br />
Figure 4-7 - Radar plot <strong>of</strong> medians for OSD<br />
This method shows similar results to <strong>the</strong> SNA method scoring higher on <strong>the</strong> validity,<br />
predictability and usefulness <strong>of</strong> <strong>the</strong> output.<br />
93
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
CDM<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
3<br />
Time to learn<br />
2<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
PN – Propositional Networks<br />
Figure 4-8 - Radar plot <strong>of</strong> medians for CDM<br />
This method scores very poorly (<strong>the</strong> worst <strong>of</strong> all <strong>the</strong> methods examined) on <strong>the</strong> ease and<br />
time to apply and learn. The method scores significantly higher on <strong>the</strong> validity,<br />
predictability and usefulness <strong>of</strong> <strong>the</strong> output.<br />
PN<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
3<br />
Time to learn<br />
2<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Figure 4-9 - Radar plot <strong>of</strong> medians for PN<br />
94
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
4.6 Discussion <strong>of</strong> Results<br />
The overall acceptance <strong>of</strong> <strong>the</strong> methodology can be understood from <strong>the</strong> last section <strong>of</strong> <strong>the</strong><br />
questionnaire. <strong>the</strong> results are shown as medians in <strong>the</strong> radar below in Figure 4-10. It can<br />
be seen that <strong>the</strong> only negative response was <strong>the</strong> time taken to complete <strong>the</strong> method. This<br />
is clearly something that all <strong>of</strong> <strong>the</strong> respondents felt strongly about with six <strong>of</strong> <strong>the</strong>m rating<br />
it as 1 and a fur<strong>the</strong>r three rating it as 2. This view is supported by <strong>the</strong> rest <strong>of</strong> <strong>the</strong> review<br />
<strong>of</strong> <strong>the</strong> method in this document.<br />
Medians<br />
Your overall acceptance <strong>of</strong> <strong>the</strong> methodology.<br />
5<br />
Please rate <strong>the</strong> overall ease <strong>of</strong> use <strong>of</strong> <strong>the</strong> methodology.<br />
4<br />
How open you perceive <strong>the</strong> methodology is to scrutiny<br />
3<br />
How w ell you think you carried out <strong>the</strong> EAST methodology<br />
2<br />
1<br />
The breadth <strong>of</strong> coverage and its ability to describe a range<br />
<strong>of</strong> behaviour.<br />
How useful do you find <strong>the</strong> EAST methodology<br />
How likely do you think <strong>the</strong> methodology is to produce<br />
consistent results<br />
Please rate <strong>the</strong> amount <strong>of</strong> resources (time & effort) required<br />
to conduct an evaluation.<br />
The extent to w hich <strong>the</strong> methodology has <strong>the</strong>oretical<br />
foundations.<br />
Figure 4-10 - Overall rating <strong>of</strong> <strong>the</strong> <strong>Methodology</strong><br />
95
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
5 Summary<br />
This report presents a review <strong>of</strong> <strong>the</strong> EAST methodology and its component methods. The<br />
following conclusions can be taken from <strong>the</strong> EAST methods review.<br />
1. The EAST methodology is an exhaustive technique. A number <strong>of</strong> different<br />
analyses are conducted and various perspectives on <strong>the</strong> scenario(s) under analysis<br />
are <strong>of</strong>fered. In its present form, <strong>the</strong> EAST methodology <strong>of</strong>fers <strong>the</strong> following<br />
analyses <strong>of</strong> a particular C4i scenario;<br />
• A step-by-step (goals, sub-goals, operations and plans) description <strong>of</strong> <strong>the</strong><br />
activity in question.<br />
• A definition <strong>of</strong> roles within <strong>the</strong> scenario.<br />
• An analysis <strong>of</strong> <strong>the</strong> agent network structure involved (e.g. network type and<br />
density).<br />
• A rating <strong>of</strong> co-ordination between agents for each team-based task step and<br />
an overall co-ordination rating.<br />
• An analysis <strong>of</strong> <strong>the</strong> current technology used during communications between<br />
agents and also recommendations for novel comms technology.<br />
• A description <strong>of</strong> <strong>the</strong> task in terms <strong>of</strong> <strong>the</strong> flow <strong>of</strong> information,<br />
communications between agents, <strong>the</strong> activity conducted by each agent<br />
involved and a timeline <strong>of</strong> activity.<br />
• An analysis <strong>of</strong> agent centrality, sociometric status and betweenness within<br />
<strong>the</strong> network involved in <strong>the</strong> scenario.<br />
• A definition <strong>of</strong> <strong>the</strong> key agents involved in <strong>the</strong> scenario.<br />
• A cognitive task analysis <strong>of</strong> operator decision making during <strong>the</strong> scenario.<br />
• A definition <strong>of</strong> <strong>the</strong> knowledge objects (information, artefacts etc) required<br />
and <strong>the</strong> knowledge objects used during <strong>the</strong> scenario.<br />
• A definition <strong>of</strong> shared knowledge or shared situation awareness during <strong>the</strong><br />
scenario.<br />
2. The EAST methodology is particularly suited to <strong>the</strong> analysis <strong>of</strong> team-based or<br />
collaborative activity, such as that seen in C4i environments. The method was<br />
originally developed for this purpose and applications so far have proved<br />
extremely successful, highlighting its suitability for such applications.<br />
3. The EAST methodology is relatively simple to use. The method requires an initial<br />
understanding <strong>of</strong> HF and experience in <strong>the</strong> application <strong>of</strong> HF methods. However,<br />
96
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
from analyst reports it can be concluded that <strong>the</strong> EAST methodology is relatively<br />
simple to apply. The methods ease <strong>of</strong> use is heightened when compared to <strong>the</strong><br />
exhaustive output that is generated from an EAST analysis.<br />
4. The EAST procedure can be made sufficiently quicker and easier with <strong>the</strong><br />
provision <strong>of</strong> s<strong>of</strong>tware assistance. The WESTT s<strong>of</strong>tware package developed during<br />
WP 1 can be used to automate a large portion <strong>of</strong> <strong>the</strong> EAST procedure (i.e.<br />
construction <strong>of</strong> OSD’s, SNA). The Agna s<strong>of</strong>tware package can also be used for<br />
<strong>the</strong> SNA part <strong>of</strong> EAST.<br />
5. The EAST methodology is generic and can be applied in any domain in which<br />
collaborative activity takes place. Despite <strong>the</strong> number <strong>of</strong> component methods<br />
involved in <strong>the</strong> EAST procedure, <strong>the</strong> method is generic and can be applied in any<br />
domain that uses collaborative activity.<br />
Despite <strong>the</strong> positive conclusions presented above, <strong>the</strong> methods review also highlighted<br />
some flaws within <strong>the</strong> EAST methodology. These are summarised below.<br />
1. Due to its exhaustive nature, <strong>the</strong> EAST methodology is time consuming to apply.<br />
A typical EAST analysis <strong>of</strong> a moderately complex scenario takes around 5 days to<br />
complete. For complex scenarios <strong>the</strong> analysis may also become laborious at times<br />
and also unwieldy in places.<br />
2. A large portion <strong>of</strong> <strong>the</strong> methods output is descriptive. A large onus is placed upon<br />
<strong>the</strong> analyst to interpret <strong>the</strong> analysis for more informative conclusions.<br />
3. In order to work properly and produce valid outputs, a large amount <strong>of</strong> access to<br />
<strong>the</strong> system and agents involved in <strong>the</strong> scenario under analysis. This is <strong>of</strong>ten<br />
difficult to obtain in domains where C4i activity takes place, due to sensitivity<br />
issues (i.e. <strong>the</strong> military, nuclear power and police domains).<br />
4. The reliability <strong>of</strong> <strong>the</strong> technique could be questionable at times. Much <strong>of</strong> <strong>the</strong><br />
analysis is based upon <strong>the</strong> subjective judgement <strong>of</strong> <strong>the</strong> analyst involved, and so<br />
intra-analyst and inter analyst reliability may suffer. However, this does not seem<br />
to have been a problem in EAST applications so far.<br />
As an overall conclusion, it is felt by <strong>the</strong> authors that <strong>the</strong> EAST methodology has been<br />
successful in its applications thus far, and despite <strong>the</strong> flaws outlined above, <strong>the</strong><br />
methodology is perfectly suited to <strong>the</strong> analysis <strong>of</strong> C4i activity. Each EAST application<br />
has produced valid and useful results which are currently forming <strong>the</strong> basis for <strong>the</strong><br />
development <strong>of</strong> a generic model <strong>of</strong> C4i. Despite its success, <strong>the</strong> EAST methodology is<br />
still evolving, and subsequently it is felt that <strong>the</strong> technique can only become more useful,<br />
usable and valid. On a final note, <strong>the</strong> following recommendations regarding <strong>the</strong><br />
development <strong>of</strong> <strong>the</strong> EAST methodology were made as a result <strong>of</strong> <strong>the</strong> methods review.<br />
1. EAST s<strong>of</strong>tware package. Despite <strong>the</strong> provision <strong>of</strong> <strong>the</strong> WESTT, AGNA and<br />
Micros<strong>of</strong>t Visio s<strong>of</strong>tware packages which automate large portions <strong>of</strong> <strong>the</strong> EAST<br />
procedure, it is felt by <strong>the</strong> authors that a comprehensive EAST s<strong>of</strong>tware package<br />
requires development. To support <strong>the</strong> EAST procedure, a s<strong>of</strong>tware package that<br />
could generate CDA, CUD, SNA, OSD, CDM and propositional network data<br />
97
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
from an initial HTA input is required. This solution is not that far away (WESTT<br />
already produces SNA, OSD and Prop network data) and it is felt that it would<br />
remove <strong>the</strong> problem <strong>of</strong> high application times and also concerns regarding <strong>the</strong><br />
reliability <strong>of</strong> <strong>the</strong> technique.<br />
2. EAST instructions booklet. There appears to be limited guidance for <strong>the</strong><br />
application <strong>of</strong> a full EAST analysis. As a solution, it is felt that an EAST<br />
guidelines or instruction booklet should be constructed, containing a step-by-step<br />
procedure and worked examples.<br />
3. EAST validation studies. Although a number <strong>of</strong> studies using <strong>the</strong> EAST<br />
methodology have already been conducted, it is felt that <strong>the</strong> methodology requires<br />
fur<strong>the</strong>r validation.<br />
4. OSD symbols. It is felt that <strong>the</strong> current set <strong>of</strong> OSD symbols are lacking when used<br />
to describe C4i activity. A set <strong>of</strong> C4i specific OSD symbols should be developed,<br />
based upon previous EAST applications.<br />
5. CUD. The technology used column in <strong>the</strong> CUD output should describe <strong>the</strong> comms<br />
technology at ei<strong>the</strong>r end (e.g. from = telephone, to = mobile phone).<br />
98
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
6 Bibliography<br />
Ainsworth, L. & Marshall, E. (1998). Issues <strong>of</strong> quality and practicality in task analysis:<br />
preliminary results from two surveys. Ergonomics 41(11), 1604-1617. Reprinted in J.<br />
Annett & N.A. Stanton (2000) op.cit. Pp. 79-89.<br />
Annett, J. 2004, Hierarchical Task <strong>Analysis</strong>. In N. A. Stanton, A. Hedge, K. Brookhuis,<br />
E. Salas, & H. Hendrick (eds.) Handbook <strong>of</strong> Human Factors methods. Taylor and<br />
Francis, London.<br />
Baber, C. & Stanton, N. A. (1996). Human error identification techniques applied to<br />
public technology: predictions compared with observed use. Applied Ergonomics, 27(2),<br />
119-131.<br />
Baber, C. and Stanton, N. A. 2004, <strong>Methodology</strong> for DTC-HFI WP1 field trials. Defence<br />
Technology Centre for Human Factors Integration, Report 2.1<br />
Baber, C., Walker, G., Stanton, N. A., & Salmon, P. (2004a). Report on Initial Trials <strong>of</strong><br />
WP1.1 <strong>Methodology</strong> conducted at fire service training college. HFI-DTC Technical<br />
Report, WP1.1.1/01, 29 th January 2004.<br />
Baber, C., Houghton, R. J., McMaster, R., Salmon, P., Stanton, N. A., Stewart, R. J., &<br />
Walker, G. (2004). Field studies in <strong>the</strong> emergency services. HFI-DTC Technical<br />
Report/WP 1.1.1/1-1, November 2004.<br />
Burke, C. S. (2004). Team Task <strong>Analysis</strong>. In N. A. Stanton, A. Hedge, K. Brookhuis, E.<br />
Salas, & H. Hendrick. (Eds.), Handbook <strong>of</strong> Human Factors methods. Boca Raton, USA,<br />
CRC Press.<br />
Dekker, A. H. (2002). C4ISR Architectures, Social Network <strong>Analysis</strong> and <strong>the</strong> FINC<br />
<strong>Methodology</strong>: An Experiment in Military Organisational Structure. DSTO Electronics<br />
and Surveillance Research Laboratory. DSTO-GD-0313<br />
Driskell, J. E. & Mullen, B. (2004) Social Network <strong>Analysis</strong>. In N. A. Stanton, A.<br />
Hedge, K. Brookhuis, E. Salas, & H. Hendrick. (2004) (Eds.) Handbook <strong>of</strong> Human<br />
Factors methods. UK, Taylor and Francis.<br />
Drury, C. G. (1990). Methods for direct observation <strong>of</strong> performance, In Wilson, J. R. And<br />
Corlett, E. N (Eds.) ‘Evaluation <strong>of</strong> Human Work – A practical ergonomics methodology’<br />
Taylor and Francis, London.<br />
Flanagan, J. C. (1954). The Critical Incident Technique. Psychological Bulletin, 51,<br />
327-358.<br />
Kirwan, B., & Ainsworth, L. K. (1992). A guide to Task <strong>Analysis</strong>, Taylor and Francis,<br />
London, UK.<br />
99
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Klein, G. & Armstrong, A. A. (2004). Critical Decision Method. In N. A. Stanton, A.<br />
Hedge, K. Brookhuis, E. Salas, & H. Hendrick. (Eds.) Handbook <strong>of</strong> Human Factors<br />
methods. UK, Taylor and Francis.<br />
Klein, G. A., Calderwood, R., & MacGregor, D. (1989). Critical Decision Method for<br />
Eliciting Knowledge. IEEE Transactions on Systems, Man and Cybernetics, 19(3), pp<br />
462-472.<br />
Marshall, A., Stanton, N., Young, M., Salmon, P., Harris, D., Demagalski, J., Waldmann,<br />
T., Dekker, S. (2003). Development <strong>of</strong> <strong>the</strong> Human Error Template – A new methodology<br />
for assessing design induced errors on aircraft flight decks. Project Report.<br />
Militello, L. G. & Hutton, J. B. (2000). Applied Cognitive Task <strong>Analysis</strong> (ACTA): A<br />
practitioner’s toolkit for understanding cognitive task demands. In J. Annett & N. S<br />
Stanton (Eds.), Task <strong>Analysis</strong>, pp 90-113. UK, Taylor & Francis.<br />
O’Hare, D., Wiggins, M., Williams, A., & Wong, W. (2000). Cognitive task analyses for<br />
decision centred design and training. In J. Annett & N. Stanton (Eds.) Task <strong>Analysis</strong>, pp<br />
170-190. UK, Taylor and Francis.<br />
Polson, P. G., Lewis, C., Rieman, J. & Wharton, C. (1992). Cognitive walkthroughs: a<br />
method for <strong>the</strong>ory based evaluation <strong>of</strong> user interfaces. International Journal <strong>of</strong> Man-<br />
Machine Studies, 36 pp741-773.<br />
Salmon, P. M., Stanton, N. A., Walker, G., & Green, D. (2004a) Human Factors Design<br />
and Evaluation Methods <strong>Review</strong>. Defence Technology Centre for Human Factors<br />
Integration, Report No. HFIDTC/WP1.3.2/1.<br />
Salmon, P. M., Stanton, N. A. and Walker, G. 2004b, National Grid Transco: Switching<br />
operations report. Defence Technology Centre for Human Factors Integration Report.<br />
Salmon, P. M., Stanton, N. A. and Walker, G. 2004c, National Grid Transco: Return to<br />
service report. Defence Technology Centre for Human Factors Integration Report.<br />
Shepherd, A. (2001). Hierarchical Task <strong>Analysis</strong>. Taylor and Francis, London.<br />
Stanton, N. A. (2004). Hierarchical Task <strong>Analysis</strong>: Developments, Applications and<br />
Extensions. HFI-DTC Technical Report.<br />
Stanton, N. A. & Stevenage, S. V. (1998). Learning to predict human error: issues <strong>of</strong><br />
acceptability, reliability and validity. Ergonomics, 41(11), 1737-1756.<br />
Stanton, N. A, and Young, M. S. (1999). A guide to methodology in ergonomics:<br />
Designing for human use, London, Taylor and Francis.<br />
Stewart, R. J., Harris, D. Stanton, N. A., Salmon, P. Linsell, M. & Dymott, R. (2004).<br />
HMS Driad: Air, Surface and Subsurface Scenario Report. Defence Technology Centre<br />
for Human Factors Integration Report.<br />
100
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Walker, G., Stanton, N. A & Young, M. (2002). Hierarchical Task <strong>Analysis</strong> <strong>of</strong> Driving<br />
(HTAoD)<br />
Walker, G. H., Stanton, N. A., Gibson, H., Baber, C., Young, M., & Green. D. (2005a).<br />
Analysing <strong>the</strong> role <strong>of</strong> communications technology in C4i scenarios: A distributed<br />
cognition approach. Accepted for publication in Special Issue <strong>of</strong> Intelligent Systems.<br />
Walker, G. H., Gibson, H., Stanton, N. A., Baber, C., Salmon, P., & Green, D. (2005b).<br />
<strong>Event</strong> analysis <strong>of</strong> systemic teamwork (EAST): A novel integration <strong>of</strong> ergonomics<br />
methods to analyse C4i activity. Accepted for publication in a Special Issue <strong>of</strong><br />
Ergonomics on C4i.<br />
Watts, L. A., & Monk, A. F. (2000). Reasoning about tasks, activities and technology to<br />
support collaboration. In J. Annett & N. Stanton (Eds.) Task <strong>Analysis</strong>. UK, Taylor and<br />
Francis.<br />
101
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
7 Appendix<br />
The fur<strong>the</strong>r <strong>the</strong> point<br />
from <strong>the</strong> centre <strong>the</strong><br />
better.<br />
Anything outside <strong>the</strong><br />
red ring is possitive<br />
inside is negative<br />
Medians<br />
5<br />
4<br />
3<br />
5<br />
5 5 5<br />
5 5<br />
4 4<br />
4 4<br />
4 4 4<br />
4 4 4<br />
4 4 4<br />
4 4 4<br />
3 3 3<br />
3 3 3 3<br />
3 3 3 3<br />
3 3 3<br />
3 3 3 3 3<br />
3<br />
3 3 3 3 3<br />
Observation<br />
HTA<br />
CDA<br />
CUD<br />
2<br />
2 2 2 2 2<br />
2<br />
2<br />
2<br />
2<br />
SNA<br />
OSD<br />
Percentage use again<br />
Observation 100%<br />
HTA 100%<br />
CDA 89%<br />
CUD 78%<br />
SNA 100%<br />
OSD 89%<br />
CDM 100%<br />
PN 100%<br />
1<br />
0<br />
Ease <strong>of</strong> learning Time to learn Ease <strong>of</strong> application Time to Apply Usefulness <strong>of</strong> output Predictability <strong>of</strong><br />
Output<br />
Validity <strong>of</strong> output<br />
CDM<br />
PN<br />
Observation<br />
HTA<br />
Ease <strong>of</strong> learning<br />
5<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
3<br />
3<br />
2<br />
2<br />
1<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
CDA<br />
CUD<br />
Ease <strong>of</strong> learning<br />
5<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
3<br />
3<br />
2<br />
2<br />
1<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
SNA<br />
OSD<br />
Ease <strong>of</strong> learning<br />
5<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
3<br />
3<br />
2<br />
2<br />
1<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
CDM<br />
PN<br />
Ease <strong>of</strong> learning<br />
5<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
3<br />
3<br />
2<br />
2<br />
1<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Figure 7-1 - <strong>Review</strong> <strong>of</strong> EAST Methods – Medians<br />
102
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Means<br />
The fur<strong>the</strong>r <strong>the</strong> point<br />
from <strong>the</strong> centre <strong>the</strong><br />
better.<br />
Anything outside <strong>the</strong><br />
red ring is possitive<br />
inside is negative<br />
Percentage use again<br />
Observation 100%<br />
HTA 100%<br />
CDA 89%<br />
CUD 78%<br />
SNA 100%<br />
OSD 89%<br />
CDM 100%<br />
PN 100%<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
4.4 4.4<br />
4.3<br />
4.2<br />
4.1 4.1<br />
4.0<br />
4.0<br />
3.9<br />
3.7<br />
3.7<br />
3.7<br />
3.6<br />
3.6 3.6<br />
3.6<br />
3.6 3.6<br />
3.3<br />
3.2<br />
3.2 3.2<br />
3.2 3.2<br />
3.1<br />
3.1<br />
3.1<br />
3.0 3.0<br />
2.9 2.9<br />
2.9<br />
2.8<br />
2.8<br />
2.8<br />
2.7<br />
2.7<br />
2.6<br />
2.4 2.4<br />
2.4<br />
2.4<br />
2.2<br />
2.2 2.2<br />
2.0<br />
1.9<br />
1.8<br />
Ease <strong>of</strong> learning Time to learn Ease <strong>of</strong> application Time to Apply Usefulness <strong>of</strong> output Predictability <strong>of</strong><br />
Output<br />
4.2 4.2<br />
4.1<br />
3.9<br />
3.7<br />
3.6<br />
3.1<br />
3.0<br />
Validity <strong>of</strong> output<br />
Observation<br />
HTA<br />
CDA<br />
CUD<br />
SNA<br />
OSD<br />
CDM<br />
PN<br />
Observation<br />
HTA<br />
Ease <strong>of</strong> learning<br />
5<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
3<br />
3<br />
2<br />
2<br />
1<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
CDA<br />
CUD<br />
Ease <strong>of</strong> learning<br />
5<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
3<br />
3<br />
2<br />
2<br />
1<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
SNA<br />
OSD<br />
Ease <strong>of</strong> learning<br />
5<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
3<br />
3<br />
2<br />
2<br />
1<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
CDM<br />
PN<br />
Ease <strong>of</strong> learning<br />
5<br />
Ease <strong>of</strong> learning<br />
5<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
4<br />
Time to learn<br />
3<br />
3<br />
2<br />
2<br />
1<br />
1<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Figure 7-2 - <strong>Review</strong> <strong>of</strong> EAST Methods – Means<br />
103
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Standard Deviations<br />
2<br />
1.7<br />
The fur<strong>the</strong>r <strong>the</strong> point from <strong>the</strong> centre <strong>the</strong> better<br />
Percentage use again<br />
Observation 100%<br />
HTA 100%<br />
CDA 89%<br />
CUD 78%<br />
SNA 100%<br />
OSD 89%<br />
CDM 100%<br />
PN 100%<br />
1.5<br />
1<br />
0.5<br />
1.4<br />
1.3<br />
1.3<br />
1.2<br />
1.1<br />
1.1<br />
1.1<br />
1.1 1.1 1.1 1.1<br />
1.0<br />
0.9<br />
0.8<br />
0.8<br />
0.6<br />
1.2<br />
1.0<br />
1.0<br />
0.9<br />
0.7<br />
0.7<br />
0.4<br />
1.3<br />
1.3<br />
1.1<br />
1.0 1.0<br />
0.8<br />
0.7<br />
0.7<br />
1.3<br />
1.2<br />
1.2<br />
1.1<br />
1.1<br />
1.1<br />
1.0<br />
0.9<br />
0.9 0.9 0.9 0.9<br />
0.7 0.7 0.7<br />
0.7<br />
1.1 1.1<br />
1.0<br />
0.9<br />
0.9<br />
0.8 0.8<br />
0.7<br />
Observation<br />
HTA<br />
CDA<br />
CUD<br />
SNA<br />
OSD<br />
CDM<br />
PN<br />
0<br />
Ease <strong>of</strong> learning Time to learn Ease <strong>of</strong> application Time to Apply Usefulness <strong>of</strong> output Predictability <strong>of</strong><br />
Output<br />
Validity <strong>of</strong> output<br />
Observation<br />
HTA<br />
Ease <strong>of</strong> learning<br />
2<br />
Ease <strong>of</strong> learning<br />
2<br />
Validity <strong>of</strong> output<br />
1.5<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
1.5<br />
Time to learn<br />
1<br />
1<br />
0.5<br />
0.5<br />
0<br />
0<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
CDA<br />
CUD<br />
Ease <strong>of</strong> learning<br />
2<br />
Ease <strong>of</strong> learning<br />
2<br />
Validity <strong>of</strong> output<br />
1.5<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
1.5<br />
Time to learn<br />
1<br />
1<br />
0.5<br />
0.5<br />
0<br />
0<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
SNA<br />
OSD<br />
Ease <strong>of</strong> learning<br />
2<br />
Ease <strong>of</strong> learning<br />
2<br />
Validity <strong>of</strong> output<br />
1.5<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
1.5<br />
Time to learn<br />
1<br />
1<br />
0.5<br />
0.5<br />
0<br />
0<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
CDM<br />
PN<br />
Ease <strong>of</strong> learning<br />
2<br />
Ease <strong>of</strong> learning<br />
2<br />
Validity <strong>of</strong> output<br />
1.5<br />
Time to learn<br />
Validity <strong>of</strong> output<br />
1.5<br />
Time to learn<br />
1<br />
1<br />
0.5<br />
0.5<br />
0<br />
0<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Predictability <strong>of</strong> Output<br />
Ease <strong>of</strong> application<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Usefulness <strong>of</strong> output<br />
Time to Apply<br />
Figure 7-3 - <strong>Review</strong> <strong>of</strong> EAST Methods – Standard Deviation<br />
104
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Medians<br />
Your overall acceptance <strong>of</strong> <strong>the</strong> methodology.<br />
5<br />
Please rate <strong>the</strong> overall ease <strong>of</strong> use <strong>of</strong> <strong>the</strong> methodology.<br />
4<br />
How open you perceive <strong>the</strong> methodology is to scrutiny<br />
3<br />
How well you think you carried out <strong>the</strong> EAST methodology<br />
2<br />
1<br />
The breadth <strong>of</strong> coverage and its ability to describe a range <strong>of</strong><br />
behaviour.<br />
How useful do you find <strong>the</strong> EAST methodology<br />
How likely do you think <strong>the</strong> methodology is to produce consistent<br />
results<br />
Please rate <strong>the</strong> amount <strong>of</strong> resources (time & effort) required to<br />
conduct an evaluation.<br />
The extent to which <strong>the</strong> methodology has <strong>the</strong>oretical foundations.<br />
Means<br />
Your overall acceptance <strong>of</strong> <strong>the</strong> methodology.<br />
5<br />
Please rate <strong>the</strong> overall ease <strong>of</strong> use <strong>of</strong> <strong>the</strong> methodology.<br />
4<br />
How open you perceive <strong>the</strong> methodology is to scrutiny<br />
3<br />
How well you think you carried out <strong>the</strong> EAST methodology<br />
2<br />
1<br />
The breadth <strong>of</strong> coverage and its ability to describe a range <strong>of</strong><br />
behaviour.<br />
How useful do you find <strong>the</strong> EAST methodology<br />
How likely do you think <strong>the</strong> methodology is to produce consistent<br />
results<br />
Please rate <strong>the</strong> amount <strong>of</strong> resources (time & effort) required to<br />
conduct an evaluation.<br />
The extent to which <strong>the</strong> methodology has <strong>the</strong>oretical foundations.<br />
Standard Deviations<br />
Your overall acceptance <strong>of</strong> <strong>the</strong> methodology.<br />
2<br />
Please rate <strong>the</strong> overall ease <strong>of</strong> use <strong>of</strong> <strong>the</strong> methodology.<br />
1.5<br />
How open you perceive <strong>the</strong> methodology is to scrutiny<br />
1<br />
How well you think you carried out <strong>the</strong> EAST methodology<br />
0.5<br />
0<br />
The breadth <strong>of</strong> coverage and its ability to describe a range <strong>of</strong><br />
behaviour.<br />
How useful do you find <strong>the</strong> EAST methodology<br />
How likely do you think <strong>the</strong> methodology is to produce consistent<br />
results<br />
Please rate <strong>the</strong> amount <strong>of</strong> resources (time & effort) required to<br />
conduct an evaluation.<br />
The extent to which <strong>the</strong> methodology has <strong>the</strong>oretical foundations.<br />
Figure 7-4 - Holistic <strong>Review</strong> <strong>of</strong> EAST<br />
105
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Figure 7-5 - EAST User Questionnaire page 1 <strong>of</strong> 4<br />
106
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Figure 7-6 - EAST User Questionnaire page 2 <strong>of</strong> 4<br />
107
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Figure 7-7 - EAST User Questionnaire page 3 <strong>of</strong> 4<br />
108
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
Figure 7-8 - EAST User Questionnaire page 4 <strong>of</strong> 4<br />
109
HFIDTC/WP1.1.3/10<br />
Version 2/ 31 October 2005<br />
- End <strong>of</strong> Document -<br />
110